edit

Machine Learning Systems Design

$$\newcommand{\tk}[1]{} \newcommand{\Amatrix}{\mathbf{A}} \newcommand{\KL}[2]{\text{KL}\left( #1\,\|\,#2 \right)} \newcommand{\Kaast}{\kernelMatrix_{\mathbf{ \ast}\mathbf{ \ast}}} \newcommand{\Kastu}{\kernelMatrix_{\mathbf{ \ast} \inducingVector}} \newcommand{\Kff}{\kernelMatrix_{\mappingFunctionVector \mappingFunctionVector}} \newcommand{\Kfu}{\kernelMatrix_{\mappingFunctionVector \inducingVector}} \newcommand{\Kuast}{\kernelMatrix_{\inducingVector \bf\ast}} \newcommand{\Kuf}{\kernelMatrix_{\inducingVector \mappingFunctionVector}} \newcommand{\Kuu}{\kernelMatrix_{\inducingVector \inducingVector}} \newcommand{\Kuui}{\Kuu^{-1}} \newcommand{\Qaast}{\mathbf{Q}_{\bf \ast \ast}} \newcommand{\Qastf}{\mathbf{Q}_{\ast \mappingFunction}} \newcommand{\Qfast}{\mathbf{Q}_{\mappingFunctionVector \bf \ast}} \newcommand{\Qff}{\mathbf{Q}_{\mappingFunctionVector \mappingFunctionVector}} \newcommand{\aMatrix}{\mathbf{A}} \newcommand{\aScalar}{a} \newcommand{\aVector}{\mathbf{a}} \newcommand{\acceleration}{a} \newcommand{\bMatrix}{\mathbf{B}} \newcommand{\bScalar}{b} \newcommand{\bVector}{\mathbf{b}} \newcommand{\basisFunc}{\phi} \newcommand{\basisFuncVector}{\boldsymbol{ \basisFunc}} \newcommand{\basisFunction}{\phi} \newcommand{\basisLocation}{\mu} \newcommand{\basisMatrix}{\boldsymbol{ \Phi}} \newcommand{\basisScalar}{\basisFunction} \newcommand{\basisVector}{\boldsymbol{ \basisFunction}} \newcommand{\activationFunction}{\phi} \newcommand{\activationMatrix}{\boldsymbol{ \Phi}} \newcommand{\activationScalar}{\basisFunction} \newcommand{\activationVector}{\boldsymbol{ \basisFunction}} \newcommand{\bigO}{\mathcal{O}} \newcommand{\binomProb}{\pi} \newcommand{\cMatrix}{\mathbf{C}} \newcommand{\cbasisMatrix}{\hat{\boldsymbol{ \Phi}}} \newcommand{\cdataMatrix}{\hat{\dataMatrix}} \newcommand{\cdataScalar}{\hat{\dataScalar}} \newcommand{\cdataVector}{\hat{\dataVector}} \newcommand{\centeredKernelMatrix}{\mathbf{ \MakeUppercase{\centeredKernelScalar}}} \newcommand{\centeredKernelScalar}{b} \newcommand{\centeredKernelVector}{\centeredKernelScalar} \newcommand{\centeringMatrix}{\mathbf{H}} \newcommand{\chiSquaredDist}[2]{\chi_{#1}^{2}\left(#2\right)} \newcommand{\chiSquaredSamp}[1]{\chi_{#1}^{2}} \newcommand{\conditionalCovariance}{\boldsymbol{ \Sigma}} \newcommand{\coregionalizationMatrix}{\mathbf{B}} \newcommand{\coregionalizationScalar}{b} \newcommand{\coregionalizationVector}{\mathbf{ \coregionalizationScalar}} \newcommand{\covDist}[2]{\text{cov}_{#2}\left(#1\right)} \newcommand{\covSamp}[1]{\text{cov}\left(#1\right)} \newcommand{\covarianceScalar}{c} \newcommand{\covarianceVector}{\mathbf{ \covarianceScalar}} \newcommand{\covarianceMatrix}{\mathbf{C}} \newcommand{\covarianceMatrixTwo}{\boldsymbol{ \Sigma}} \newcommand{\croupierScalar}{s} \newcommand{\croupierVector}{\mathbf{ \croupierScalar}} \newcommand{\croupierMatrix}{\mathbf{ \MakeUppercase{\croupierScalar}}} \newcommand{\dataDim}{p} \newcommand{\dataIndex}{i} \newcommand{\dataIndexTwo}{j} \newcommand{\dataMatrix}{\mathbf{Y}} \newcommand{\dataScalar}{y} \newcommand{\dataSet}{\mathcal{D}} \newcommand{\dataStd}{\sigma} \newcommand{\dataVector}{\mathbf{ \dataScalar}} \newcommand{\decayRate}{d} \newcommand{\degreeMatrix}{\mathbf{ \MakeUppercase{\degreeScalar}}} \newcommand{\degreeScalar}{d} \newcommand{\degreeVector}{\mathbf{ \degreeScalar}} \newcommand{\diag}[1]{\text{diag}\left(#1\right)} \newcommand{\diagonalMatrix}{\mathbf{D}} \newcommand{\diff}[2]{\frac{\text{d}#1}{\text{d}#2}} \newcommand{\diffTwo}[2]{\frac{\text{d}^2#1}{\text{d}#2^2}} \newcommand{\displacement}{x} \newcommand{\displacementVector}{\textbf{\displacement}} \newcommand{\distanceMatrix}{\mathbf{ \MakeUppercase{\distanceScalar}}} \newcommand{\distanceScalar}{d} \newcommand{\distanceVector}{\mathbf{ \distanceScalar}} \newcommand{\eigenvaltwo}{\ell} \newcommand{\eigenvaltwoMatrix}{\mathbf{L}} \newcommand{\eigenvaltwoVector}{\mathbf{l}} \newcommand{\eigenvalue}{\lambda} \newcommand{\eigenvalueMatrix}{\boldsymbol{ \Lambda}} \newcommand{\eigenvalueVector}{\boldsymbol{ \lambda}} \newcommand{\eigenvector}{\mathbf{ \eigenvectorScalar}} \newcommand{\eigenvectorMatrix}{\mathbf{U}} \newcommand{\eigenvectorScalar}{u} \newcommand{\eigenvectwo}{\mathbf{v}} \newcommand{\eigenvectwoMatrix}{\mathbf{V}} \newcommand{\eigenvectwoScalar}{v} \newcommand{\entropy}[1]{\mathcal{H}\left(#1\right)} \newcommand{\errorFunction}{E} \newcommand{\expDist}[2]{\left\langle#1\right\rangle_{#2}} \newcommand{\expSamp}[1]{\left\langle#1\right\rangle} \newcommand{\expectation}[1]{\left\langle #1 \right\rangle } \newcommand{\expectationDist}[2]{\left\langle #1 \right\rangle _{#2}} \newcommand{\expectedDistanceMatrix}{\mathcal{D}} \newcommand{\eye}{\mathbf{I}} \newcommand{\fantasyDim}{r} \newcommand{\fantasyMatrix}{\mathbf{ \MakeUppercase{\fantasyScalar}}} \newcommand{\fantasyScalar}{z} \newcommand{\fantasyVector}{\mathbf{ \fantasyScalar}} \newcommand{\featureStd}{\varsigma} \newcommand{\gammaCdf}[3]{\mathcal{GAMMA CDF}\left(#1|#2,#3\right)} \newcommand{\gammaDist}[3]{\mathcal{G}\left(#1|#2,#3\right)} \newcommand{\gammaSamp}[2]{\mathcal{G}\left(#1,#2\right)} \newcommand{\gaussianDist}[3]{\mathcal{N}\left(#1|#2,#3\right)} \newcommand{\gaussianSamp}[2]{\mathcal{N}\left(#1,#2\right)} \newcommand{\uniformDist}[3]{\mathcal{U}\left(#1|#2,#3\right)} \newcommand{\uniformSamp}[2]{\mathcal{U}\left(#1,#2\right)} \newcommand{\given}{|} \newcommand{\half}{\frac{1}{2}} \newcommand{\heaviside}{H} \newcommand{\hiddenMatrix}{\mathbf{ \MakeUppercase{\hiddenScalar}}} \newcommand{\hiddenScalar}{h} \newcommand{\hiddenVector}{\mathbf{ \hiddenScalar}} \newcommand{\identityMatrix}{\eye} \newcommand{\inducingInputScalar}{z} \newcommand{\inducingInputVector}{\mathbf{ \inducingInputScalar}} \newcommand{\inducingInputMatrix}{\mathbf{Z}} \newcommand{\inducingScalar}{u} \newcommand{\inducingVector}{\mathbf{ \inducingScalar}} \newcommand{\inducingMatrix}{\mathbf{U}} \newcommand{\inlineDiff}[2]{\text{d}#1/\text{d}#2} \newcommand{\inputDim}{q} \newcommand{\inputMatrix}{\mathbf{X}} \newcommand{\inputScalar}{x} \newcommand{\inputSpace}{\mathcal{X}} \newcommand{\inputVals}{\inputVector} \newcommand{\inputVector}{\mathbf{ \inputScalar}} \newcommand{\iterNum}{k} \newcommand{\kernel}{\kernelScalar} \newcommand{\kernelMatrix}{\mathbf{K}} \newcommand{\kernelScalar}{k} \newcommand{\kernelVector}{\mathbf{ \kernelScalar}} \newcommand{\kff}{\kernelScalar_{\mappingFunction \mappingFunction}} \newcommand{\kfu}{\kernelVector_{\mappingFunction \inducingScalar}} \newcommand{\kuf}{\kernelVector_{\inducingScalar \mappingFunction}} \newcommand{\kuu}{\kernelVector_{\inducingScalar \inducingScalar}} \newcommand{\lagrangeMultiplier}{\lambda} \newcommand{\lagrangeMultiplierMatrix}{\boldsymbol{ \Lambda}} \newcommand{\lagrangian}{L} \newcommand{\laplacianFactor}{\mathbf{ \MakeUppercase{\laplacianFactorScalar}}} \newcommand{\laplacianFactorScalar}{m} \newcommand{\laplacianFactorVector}{\mathbf{ \laplacianFactorScalar}} \newcommand{\laplacianMatrix}{\mathbf{L}} \newcommand{\laplacianScalar}{\ell} \newcommand{\laplacianVector}{\mathbf{ \ell}} \newcommand{\latentDim}{q} \newcommand{\latentDistanceMatrix}{\boldsymbol{ \Delta}} \newcommand{\latentDistanceScalar}{\delta} \newcommand{\latentDistanceVector}{\boldsymbol{ \delta}} \newcommand{\latentForce}{f} \newcommand{\latentFunction}{u} \newcommand{\latentFunctionVector}{\mathbf{ \latentFunction}} \newcommand{\latentFunctionMatrix}{\mathbf{ \MakeUppercase{\latentFunction}}} \newcommand{\latentIndex}{j} \newcommand{\latentScalar}{z} \newcommand{\latentVector}{\mathbf{ \latentScalar}} \newcommand{\latentMatrix}{\mathbf{Z}} \newcommand{\learnRate}{\eta} \newcommand{\lengthScale}{\ell} \newcommand{\rbfWidth}{\ell} \newcommand{\likelihoodBound}{\mathcal{L}} \newcommand{\likelihoodFunction}{L} \newcommand{\locationScalar}{\mu} \newcommand{\locationVector}{\boldsymbol{ \locationScalar}} \newcommand{\locationMatrix}{\mathbf{M}} \newcommand{\variance}[1]{\text{var}\left( #1 \right)} \newcommand{\mappingFunction}{f} \newcommand{\mappingFunctionMatrix}{\mathbf{F}} \newcommand{\mappingFunctionTwo}{g} \newcommand{\mappingFunctionTwoMatrix}{\mathbf{G}} \newcommand{\mappingFunctionTwoVector}{\mathbf{ \mappingFunctionTwo}} \newcommand{\mappingFunctionVector}{\mathbf{ \mappingFunction}} \newcommand{\scaleScalar}{s} \newcommand{\mappingScalar}{w} \newcommand{\mappingVector}{\mathbf{ \mappingScalar}} \newcommand{\mappingMatrix}{\mathbf{W}} \newcommand{\mappingScalarTwo}{v} \newcommand{\mappingVectorTwo}{\mathbf{ \mappingScalarTwo}} \newcommand{\mappingMatrixTwo}{\mathbf{V}} \newcommand{\maxIters}{K} \newcommand{\meanMatrix}{\mathbf{M}} \newcommand{\meanScalar}{\mu} \newcommand{\meanTwoMatrix}{\mathbf{M}} \newcommand{\meanTwoScalar}{m} \newcommand{\meanTwoVector}{\mathbf{ \meanTwoScalar}} \newcommand{\meanVector}{\boldsymbol{ \meanScalar}} \newcommand{\mrnaConcentration}{m} \newcommand{\naturalFrequency}{\omega} \newcommand{\neighborhood}[1]{\mathcal{N}\left( #1 \right)} \newcommand{\neilurl}{http://inverseprobability.com/} \newcommand{\noiseMatrix}{\boldsymbol{ E}} \newcommand{\noiseScalar}{\epsilon} \newcommand{\noiseVector}{\boldsymbol{ \epsilon}} \newcommand{\noiseStd}{\sigma} \newcommand{\norm}[1]{\left\Vert #1 \right\Vert} \newcommand{\normalizedLaplacianMatrix}{\hat{\mathbf{L}}} \newcommand{\normalizedLaplacianScalar}{\hat{\ell}} \newcommand{\normalizedLaplacianVector}{\hat{\mathbf{ \ell}}} \newcommand{\numActive}{m} \newcommand{\numBasisFunc}{m} \newcommand{\numComponents}{m} \newcommand{\numComps}{K} \newcommand{\numData}{n} \newcommand{\numFeatures}{K} \newcommand{\numHidden}{h} \newcommand{\numInducing}{m} \newcommand{\numLayers}{\ell} \newcommand{\numNeighbors}{K} \newcommand{\numSequences}{s} \newcommand{\numSuccess}{s} \newcommand{\numTasks}{m} \newcommand{\numTime}{T} \newcommand{\numTrials}{S} \newcommand{\outputIndex}{j} \newcommand{\paramVector}{\boldsymbol{ \theta}} \newcommand{\parameterMatrix}{\boldsymbol{ \Theta}} \newcommand{\parameterScalar}{\theta} \newcommand{\parameterVector}{\boldsymbol{ \parameterScalar}} \newcommand{\partDiff}[2]{\frac{\partial#1}{\partial#2}} \newcommand{\precisionScalar}{j} \newcommand{\precisionVector}{\mathbf{ \precisionScalar}} \newcommand{\precisionMatrix}{\mathbf{J}} \newcommand{\pseudotargetScalar}{\widetilde{y}} \newcommand{\pseudotargetVector}{\mathbf{ \pseudotargetScalar}} \newcommand{\pseudotargetMatrix}{\mathbf{ \widetilde{Y}}} \newcommand{\rank}[1]{\text{rank}\left(#1\right)} \newcommand{\rayleighDist}[2]{\mathcal{R}\left(#1|#2\right)} \newcommand{\rayleighSamp}[1]{\mathcal{R}\left(#1\right)} \newcommand{\responsibility}{r} \newcommand{\rotationScalar}{r} \newcommand{\rotationVector}{\mathbf{ \rotationScalar}} \newcommand{\rotationMatrix}{\mathbf{R}} \newcommand{\sampleCovScalar}{s} \newcommand{\sampleCovVector}{\mathbf{ \sampleCovScalar}} \newcommand{\sampleCovMatrix}{\mathbf{s}} \newcommand{\scalarProduct}[2]{\left\langle{#1},{#2}\right\rangle} \newcommand{\sign}[1]{\text{sign}\left(#1\right)} \newcommand{\sigmoid}[1]{\sigma\left(#1\right)} \newcommand{\singularvalue}{\ell} \newcommand{\singularvalueMatrix}{\mathbf{L}} \newcommand{\singularvalueVector}{\mathbf{l}} \newcommand{\sorth}{\mathbf{u}} \newcommand{\spar}{\lambda} \newcommand{\trace}[1]{\text{tr}\left(#1\right)} \newcommand{\BasalRate}{B} \newcommand{\DampingCoefficient}{C} \newcommand{\DecayRate}{D} \newcommand{\Displacement}{X} \newcommand{\LatentForce}{F} \newcommand{\Mass}{M} \newcommand{\Sensitivity}{S} \newcommand{\basalRate}{b} \newcommand{\dampingCoefficient}{c} \newcommand{\mass}{m} \newcommand{\sensitivity}{s} \newcommand{\springScalar}{\kappa} \newcommand{\springVector}{\boldsymbol{ \kappa}} \newcommand{\springMatrix}{\boldsymbol{ \mathcal{K}}} \newcommand{\tfConcentration}{p} \newcommand{\tfDecayRate}{\delta} \newcommand{\tfMrnaConcentration}{f} \newcommand{\tfVector}{\mathbf{ \tfConcentration}} \newcommand{\velocity}{v} \newcommand{\sufficientStatsScalar}{g} \newcommand{\sufficientStatsVector}{\mathbf{ \sufficientStatsScalar}} \newcommand{\sufficientStatsMatrix}{\mathbf{G}} \newcommand{\switchScalar}{s} \newcommand{\switchVector}{\mathbf{ \switchScalar}} \newcommand{\switchMatrix}{\mathbf{S}} \newcommand{\tr}[1]{\text{tr}\left(#1\right)} \newcommand{\loneNorm}[1]{\left\Vert #1 \right\Vert_1} \newcommand{\ltwoNorm}[1]{\left\Vert #1 \right\Vert_2} \newcommand{\onenorm}[1]{\left\vert#1\right\vert_1} \newcommand{\twonorm}[1]{\left\Vert #1 \right\Vert} \newcommand{\vScalar}{v} \newcommand{\vVector}{\mathbf{v}} \newcommand{\vMatrix}{\mathbf{V}} \newcommand{\varianceDist}[2]{\text{var}_{#2}\left( #1 \right)} \newcommand{\vecb}[1]{\left(#1\right):} \newcommand{\weightScalar}{w} \newcommand{\weightVector}{\mathbf{ \weightScalar}} \newcommand{\weightMatrix}{\mathbf{W}} \newcommand{\weightedAdjacencyMatrix}{\mathbf{A}} \newcommand{\weightedAdjacencyScalar}{a} \newcommand{\weightedAdjacencyVector}{\mathbf{ \weightedAdjacencyScalar}} \newcommand{\onesVector}{\mathbf{1}} \newcommand{\zerosVector}{\mathbf{0}} $$
at Cambridge AI Group Seminar on Nov 5, 2019 [reveal]
Neil D. Lawrence, University of Sheffield

Abstract

Machine learning solutions, in particular those based on deep learning methods, form an underpinning of the current revolution in “artificial intelligence” that has dominated popular press headlines and is having a significant influence on the wider tech agenda. In this talk I will give an overview of where we are now with machine learning solutions, and what challenges we face both in the near and far future. These include practical application of existing algorithms in the face of the need to explain decision-making, mechanisms for improving the quality and availability of data, dealing with large unstructured datasets.

$$\newcommand{\tk}[1]{} \newcommand{\Amatrix}{\mathbf{A}} \newcommand{\KL}[2]{\text{KL}\left( #1\,\|\,#2 \right)} \newcommand{\Kaast}{\kernelMatrix_{\mathbf{ \ast}\mathbf{ \ast}}} \newcommand{\Kastu}{\kernelMatrix_{\mathbf{ \ast} \inducingVector}} \newcommand{\Kff}{\kernelMatrix_{\mappingFunctionVector \mappingFunctionVector}} \newcommand{\Kfu}{\kernelMatrix_{\mappingFunctionVector \inducingVector}} \newcommand{\Kuast}{\kernelMatrix_{\inducingVector \bf\ast}} \newcommand{\Kuf}{\kernelMatrix_{\inducingVector \mappingFunctionVector}} \newcommand{\Kuu}{\kernelMatrix_{\inducingVector \inducingVector}} \newcommand{\Kuui}{\Kuu^{-1}} \newcommand{\Qaast}{\mathbf{Q}_{\bf \ast \ast}} \newcommand{\Qastf}{\mathbf{Q}_{\ast \mappingFunction}} \newcommand{\Qfast}{\mathbf{Q}_{\mappingFunctionVector \bf \ast}} \newcommand{\Qff}{\mathbf{Q}_{\mappingFunctionVector \mappingFunctionVector}} \newcommand{\aMatrix}{\mathbf{A}} \newcommand{\aScalar}{a} \newcommand{\aVector}{\mathbf{a}} \newcommand{\acceleration}{a} \newcommand{\bMatrix}{\mathbf{B}} \newcommand{\bScalar}{b} \newcommand{\bVector}{\mathbf{b}} \newcommand{\basisFunc}{\phi} \newcommand{\basisFuncVector}{\boldsymbol{ \basisFunc}} \newcommand{\basisFunction}{\phi} \newcommand{\basisLocation}{\mu} \newcommand{\basisMatrix}{\boldsymbol{ \Phi}} \newcommand{\basisScalar}{\basisFunction} \newcommand{\basisVector}{\boldsymbol{ \basisFunction}} \newcommand{\activationFunction}{\phi} \newcommand{\activationMatrix}{\boldsymbol{ \Phi}} \newcommand{\activationScalar}{\basisFunction} \newcommand{\activationVector}{\boldsymbol{ \basisFunction}} \newcommand{\bigO}{\mathcal{O}} \newcommand{\binomProb}{\pi} \newcommand{\cMatrix}{\mathbf{C}} \newcommand{\cbasisMatrix}{\hat{\boldsymbol{ \Phi}}} \newcommand{\cdataMatrix}{\hat{\dataMatrix}} \newcommand{\cdataScalar}{\hat{\dataScalar}} \newcommand{\cdataVector}{\hat{\dataVector}} \newcommand{\centeredKernelMatrix}{\mathbf{ \MakeUppercase{\centeredKernelScalar}}} \newcommand{\centeredKernelScalar}{b} \newcommand{\centeredKernelVector}{\centeredKernelScalar} \newcommand{\centeringMatrix}{\mathbf{H}} \newcommand{\chiSquaredDist}[2]{\chi_{#1}^{2}\left(#2\right)} \newcommand{\chiSquaredSamp}[1]{\chi_{#1}^{2}} \newcommand{\conditionalCovariance}{\boldsymbol{ \Sigma}} \newcommand{\coregionalizationMatrix}{\mathbf{B}} \newcommand{\coregionalizationScalar}{b} \newcommand{\coregionalizationVector}{\mathbf{ \coregionalizationScalar}} \newcommand{\covDist}[2]{\text{cov}_{#2}\left(#1\right)} \newcommand{\covSamp}[1]{\text{cov}\left(#1\right)} \newcommand{\covarianceScalar}{c} \newcommand{\covarianceVector}{\mathbf{ \covarianceScalar}} \newcommand{\covarianceMatrix}{\mathbf{C}} \newcommand{\covarianceMatrixTwo}{\boldsymbol{ \Sigma}} \newcommand{\croupierScalar}{s} \newcommand{\croupierVector}{\mathbf{ \croupierScalar}} \newcommand{\croupierMatrix}{\mathbf{ \MakeUppercase{\croupierScalar}}} \newcommand{\dataDim}{p} \newcommand{\dataIndex}{i} \newcommand{\dataIndexTwo}{j} \newcommand{\dataMatrix}{\mathbf{Y}} \newcommand{\dataScalar}{y} \newcommand{\dataSet}{\mathcal{D}} \newcommand{\dataStd}{\sigma} \newcommand{\dataVector}{\mathbf{ \dataScalar}} \newcommand{\decayRate}{d} \newcommand{\degreeMatrix}{\mathbf{ \MakeUppercase{\degreeScalar}}} \newcommand{\degreeScalar}{d} \newcommand{\degreeVector}{\mathbf{ \degreeScalar}} \newcommand{\diag}[1]{\text{diag}\left(#1\right)} \newcommand{\diagonalMatrix}{\mathbf{D}} \newcommand{\diff}[2]{\frac{\text{d}#1}{\text{d}#2}} \newcommand{\diffTwo}[2]{\frac{\text{d}^2#1}{\text{d}#2^2}} \newcommand{\displacement}{x} \newcommand{\displacementVector}{\textbf{\displacement}} \newcommand{\distanceMatrix}{\mathbf{ \MakeUppercase{\distanceScalar}}} \newcommand{\distanceScalar}{d} \newcommand{\distanceVector}{\mathbf{ \distanceScalar}} \newcommand{\eigenvaltwo}{\ell} \newcommand{\eigenvaltwoMatrix}{\mathbf{L}} \newcommand{\eigenvaltwoVector}{\mathbf{l}} \newcommand{\eigenvalue}{\lambda} \newcommand{\eigenvalueMatrix}{\boldsymbol{ \Lambda}} \newcommand{\eigenvalueVector}{\boldsymbol{ \lambda}} \newcommand{\eigenvector}{\mathbf{ \eigenvectorScalar}} \newcommand{\eigenvectorMatrix}{\mathbf{U}} \newcommand{\eigenvectorScalar}{u} \newcommand{\eigenvectwo}{\mathbf{v}} \newcommand{\eigenvectwoMatrix}{\mathbf{V}} \newcommand{\eigenvectwoScalar}{v} \newcommand{\entropy}[1]{\mathcal{H}\left(#1\right)} \newcommand{\errorFunction}{E} \newcommand{\expDist}[2]{\left\langle#1\right\rangle_{#2}} \newcommand{\expSamp}[1]{\left\langle#1\right\rangle} \newcommand{\expectation}[1]{\left\langle #1 \right\rangle } \newcommand{\expectationDist}[2]{\left\langle #1 \right\rangle _{#2}} \newcommand{\expectedDistanceMatrix}{\mathcal{D}} \newcommand{\eye}{\mathbf{I}} \newcommand{\fantasyDim}{r} \newcommand{\fantasyMatrix}{\mathbf{ \MakeUppercase{\fantasyScalar}}} \newcommand{\fantasyScalar}{z} \newcommand{\fantasyVector}{\mathbf{ \fantasyScalar}} \newcommand{\featureStd}{\varsigma} \newcommand{\gammaCdf}[3]{\mathcal{GAMMA CDF}\left(#1|#2,#3\right)} \newcommand{\gammaDist}[3]{\mathcal{G}\left(#1|#2,#3\right)} \newcommand{\gammaSamp}[2]{\mathcal{G}\left(#1,#2\right)} \newcommand{\gaussianDist}[3]{\mathcal{N}\left(#1|#2,#3\right)} \newcommand{\gaussianSamp}[2]{\mathcal{N}\left(#1,#2\right)} \newcommand{\uniformDist}[3]{\mathcal{U}\left(#1|#2,#3\right)} \newcommand{\uniformSamp}[2]{\mathcal{U}\left(#1,#2\right)} \newcommand{\given}{|} \newcommand{\half}{\frac{1}{2}} \newcommand{\heaviside}{H} \newcommand{\hiddenMatrix}{\mathbf{ \MakeUppercase{\hiddenScalar}}} \newcommand{\hiddenScalar}{h} \newcommand{\hiddenVector}{\mathbf{ \hiddenScalar}} \newcommand{\identityMatrix}{\eye} \newcommand{\inducingInputScalar}{z} \newcommand{\inducingInputVector}{\mathbf{ \inducingInputScalar}} \newcommand{\inducingInputMatrix}{\mathbf{Z}} \newcommand{\inducingScalar}{u} \newcommand{\inducingVector}{\mathbf{ \inducingScalar}} \newcommand{\inducingMatrix}{\mathbf{U}} \newcommand{\inlineDiff}[2]{\text{d}#1/\text{d}#2} \newcommand{\inputDim}{q} \newcommand{\inputMatrix}{\mathbf{X}} \newcommand{\inputScalar}{x} \newcommand{\inputSpace}{\mathcal{X}} \newcommand{\inputVals}{\inputVector} \newcommand{\inputVector}{\mathbf{ \inputScalar}} \newcommand{\iterNum}{k} \newcommand{\kernel}{\kernelScalar} \newcommand{\kernelMatrix}{\mathbf{K}} \newcommand{\kernelScalar}{k} \newcommand{\kernelVector}{\mathbf{ \kernelScalar}} \newcommand{\kff}{\kernelScalar_{\mappingFunction \mappingFunction}} \newcommand{\kfu}{\kernelVector_{\mappingFunction \inducingScalar}} \newcommand{\kuf}{\kernelVector_{\inducingScalar \mappingFunction}} \newcommand{\kuu}{\kernelVector_{\inducingScalar \inducingScalar}} \newcommand{\lagrangeMultiplier}{\lambda} \newcommand{\lagrangeMultiplierMatrix}{\boldsymbol{ \Lambda}} \newcommand{\lagrangian}{L} \newcommand{\laplacianFactor}{\mathbf{ \MakeUppercase{\laplacianFactorScalar}}} \newcommand{\laplacianFactorScalar}{m} \newcommand{\laplacianFactorVector}{\mathbf{ \laplacianFactorScalar}} \newcommand{\laplacianMatrix}{\mathbf{L}} \newcommand{\laplacianScalar}{\ell} \newcommand{\laplacianVector}{\mathbf{ \ell}} \newcommand{\latentDim}{q} \newcommand{\latentDistanceMatrix}{\boldsymbol{ \Delta}} \newcommand{\latentDistanceScalar}{\delta} \newcommand{\latentDistanceVector}{\boldsymbol{ \delta}} \newcommand{\latentForce}{f} \newcommand{\latentFunction}{u} \newcommand{\latentFunctionVector}{\mathbf{ \latentFunction}} \newcommand{\latentFunctionMatrix}{\mathbf{ \MakeUppercase{\latentFunction}}} \newcommand{\latentIndex}{j} \newcommand{\latentScalar}{z} \newcommand{\latentVector}{\mathbf{ \latentScalar}} \newcommand{\latentMatrix}{\mathbf{Z}} \newcommand{\learnRate}{\eta} \newcommand{\lengthScale}{\ell} \newcommand{\rbfWidth}{\ell} \newcommand{\likelihoodBound}{\mathcal{L}} \newcommand{\likelihoodFunction}{L} \newcommand{\locationScalar}{\mu} \newcommand{\locationVector}{\boldsymbol{ \locationScalar}} \newcommand{\locationMatrix}{\mathbf{M}} \newcommand{\variance}[1]{\text{var}\left( #1 \right)} \newcommand{\mappingFunction}{f} \newcommand{\mappingFunctionMatrix}{\mathbf{F}} \newcommand{\mappingFunctionTwo}{g} \newcommand{\mappingFunctionTwoMatrix}{\mathbf{G}} \newcommand{\mappingFunctionTwoVector}{\mathbf{ \mappingFunctionTwo}} \newcommand{\mappingFunctionVector}{\mathbf{ \mappingFunction}} \newcommand{\scaleScalar}{s} \newcommand{\mappingScalar}{w} \newcommand{\mappingVector}{\mathbf{ \mappingScalar}} \newcommand{\mappingMatrix}{\mathbf{W}} \newcommand{\mappingScalarTwo}{v} \newcommand{\mappingVectorTwo}{\mathbf{ \mappingScalarTwo}} \newcommand{\mappingMatrixTwo}{\mathbf{V}} \newcommand{\maxIters}{K} \newcommand{\meanMatrix}{\mathbf{M}} \newcommand{\meanScalar}{\mu} \newcommand{\meanTwoMatrix}{\mathbf{M}} \newcommand{\meanTwoScalar}{m} \newcommand{\meanTwoVector}{\mathbf{ \meanTwoScalar}} \newcommand{\meanVector}{\boldsymbol{ \meanScalar}} \newcommand{\mrnaConcentration}{m} \newcommand{\naturalFrequency}{\omega} \newcommand{\neighborhood}[1]{\mathcal{N}\left( #1 \right)} \newcommand{\neilurl}{http://inverseprobability.com/} \newcommand{\noiseMatrix}{\boldsymbol{ E}} \newcommand{\noiseScalar}{\epsilon} \newcommand{\noiseVector}{\boldsymbol{ \epsilon}} \newcommand{\noiseStd}{\sigma} \newcommand{\norm}[1]{\left\Vert #1 \right\Vert} \newcommand{\normalizedLaplacianMatrix}{\hat{\mathbf{L}}} \newcommand{\normalizedLaplacianScalar}{\hat{\ell}} \newcommand{\normalizedLaplacianVector}{\hat{\mathbf{ \ell}}} \newcommand{\numActive}{m} \newcommand{\numBasisFunc}{m} \newcommand{\numComponents}{m} \newcommand{\numComps}{K} \newcommand{\numData}{n} \newcommand{\numFeatures}{K} \newcommand{\numHidden}{h} \newcommand{\numInducing}{m} \newcommand{\numLayers}{\ell} \newcommand{\numNeighbors}{K} \newcommand{\numSequences}{s} \newcommand{\numSuccess}{s} \newcommand{\numTasks}{m} \newcommand{\numTime}{T} \newcommand{\numTrials}{S} \newcommand{\outputIndex}{j} \newcommand{\paramVector}{\boldsymbol{ \theta}} \newcommand{\parameterMatrix}{\boldsymbol{ \Theta}} \newcommand{\parameterScalar}{\theta} \newcommand{\parameterVector}{\boldsymbol{ \parameterScalar}} \newcommand{\partDiff}[2]{\frac{\partial#1}{\partial#2}} \newcommand{\precisionScalar}{j} \newcommand{\precisionVector}{\mathbf{ \precisionScalar}} \newcommand{\precisionMatrix}{\mathbf{J}} \newcommand{\pseudotargetScalar}{\widetilde{y}} \newcommand{\pseudotargetVector}{\mathbf{ \pseudotargetScalar}} \newcommand{\pseudotargetMatrix}{\mathbf{ \widetilde{Y}}} \newcommand{\rank}[1]{\text{rank}\left(#1\right)} \newcommand{\rayleighDist}[2]{\mathcal{R}\left(#1|#2\right)} \newcommand{\rayleighSamp}[1]{\mathcal{R}\left(#1\right)} \newcommand{\responsibility}{r} \newcommand{\rotationScalar}{r} \newcommand{\rotationVector}{\mathbf{ \rotationScalar}} \newcommand{\rotationMatrix}{\mathbf{R}} \newcommand{\sampleCovScalar}{s} \newcommand{\sampleCovVector}{\mathbf{ \sampleCovScalar}} \newcommand{\sampleCovMatrix}{\mathbf{s}} \newcommand{\scalarProduct}[2]{\left\langle{#1},{#2}\right\rangle} \newcommand{\sign}[1]{\text{sign}\left(#1\right)} \newcommand{\sigmoid}[1]{\sigma\left(#1\right)} \newcommand{\singularvalue}{\ell} \newcommand{\singularvalueMatrix}{\mathbf{L}} \newcommand{\singularvalueVector}{\mathbf{l}} \newcommand{\sorth}{\mathbf{u}} \newcommand{\spar}{\lambda} \newcommand{\trace}[1]{\text{tr}\left(#1\right)} \newcommand{\BasalRate}{B} \newcommand{\DampingCoefficient}{C} \newcommand{\DecayRate}{D} \newcommand{\Displacement}{X} \newcommand{\LatentForce}{F} \newcommand{\Mass}{M} \newcommand{\Sensitivity}{S} \newcommand{\basalRate}{b} \newcommand{\dampingCoefficient}{c} \newcommand{\mass}{m} \newcommand{\sensitivity}{s} \newcommand{\springScalar}{\kappa} \newcommand{\springVector}{\boldsymbol{ \kappa}} \newcommand{\springMatrix}{\boldsymbol{ \mathcal{K}}} \newcommand{\tfConcentration}{p} \newcommand{\tfDecayRate}{\delta} \newcommand{\tfMrnaConcentration}{f} \newcommand{\tfVector}{\mathbf{ \tfConcentration}} \newcommand{\velocity}{v} \newcommand{\sufficientStatsScalar}{g} \newcommand{\sufficientStatsVector}{\mathbf{ \sufficientStatsScalar}} \newcommand{\sufficientStatsMatrix}{\mathbf{G}} \newcommand{\switchScalar}{s} \newcommand{\switchVector}{\mathbf{ \switchScalar}} \newcommand{\switchMatrix}{\mathbf{S}} \newcommand{\tr}[1]{\text{tr}\left(#1\right)} \newcommand{\loneNorm}[1]{\left\Vert #1 \right\Vert_1} \newcommand{\ltwoNorm}[1]{\left\Vert #1 \right\Vert_2} \newcommand{\onenorm}[1]{\left\vert#1\right\vert_1} \newcommand{\twonorm}[1]{\left\Vert #1 \right\Vert} \newcommand{\vScalar}{v} \newcommand{\vVector}{\mathbf{v}} \newcommand{\vMatrix}{\mathbf{V}} \newcommand{\varianceDist}[2]{\text{var}_{#2}\left( #1 \right)} \newcommand{\vecb}[1]{\left(#1\right):} \newcommand{\weightScalar}{w} \newcommand{\weightVector}{\mathbf{ \weightScalar}} \newcommand{\weightMatrix}{\mathbf{W}} \newcommand{\weightedAdjacencyMatrix}{\mathbf{A}} \newcommand{\weightedAdjacencyScalar}{a} \newcommand{\weightedAdjacencyVector}{\mathbf{ \weightedAdjacencyScalar}} \newcommand{\onesVector}{\mathbf{1}} \newcommand{\zerosVector}{\mathbf{0}} $$

Figure: Some software components in a ride allocation system. Circled components are hypothetical, rectangles represent actual data.

The Promise of AI [edit]

The promise of the fourth industrial revolution is that this wave of automation is the first wave of automation where the machines adapt to serve us rather than us adapting to serve the machine.

That promise will remain unfulfilled with our current approach to systems design.

Turing AI Fellowship [edit]

From December 2019 I begin a Senior AI Fellowship at the Turing Institute funded by the Office for AI to investigate the consequences of deploying complex AI systems.

The notion relates from the “Promise of AI”: it promises to be the first generation of automation technology that will adapt to us, rather than us adapting to it. The premise of the project is that this promise will remain unfulfilled with current approaches to systems design and deployment.

Project Description

It used to be true that computers only did what we programmed them to do, but today AI systems are learning from our data. This introduces new problems in how these systems respond to their environment.

We need to better monitor how data is influencing decision making and take corrective action as required.

Aim

Our aim is to scale our ability to deploy safe and reliable AI solutions. Our technical approach is to do this through data-oriented software engineering practices and deep system emulation. We will do this through a significant extension of the notion of Automated ML (AutoML) to Automated AI (AutoAI), this relies on a shift from Bayesian Optimisation to Bayesian System Optimisation. The project will develop a toolkit for automating the deployment, maintenance and monitoring of artificial intelligence systems.

Motivating Examples

SafeBoda [edit]

Figure: SafeBoda is a ride allocation system for Boda Boda drivers. Let’s imagine the capabilities we need for such an AI system.

SafeBoda is a Kampala based rider allocation system for Boda Boda drivers. Boda boda are motorcycle taxis which give employment to, often young men, across Kampala. Safe Boda is driven by the knowledge that road accidents are set to match HIV/AIDS as the highest cause of death in low/middle income families by 2030.

With road accidents set to match HIV/AIDS as the highest cause of death in low/middle income countries by 2030, SafeBoda’s aim is to modernise informal transportation and ensure safe access to mobility.

SafeBoda and other projects like Kudu provide us with our motivating examples. Our aim is to create an ecosystem for machine learing system deployment that minimises the operational load. Ideally, we would like complex AI systems to be maintainable by a small team, e.g. two people, with Masters-level education from the institutions that host Data Science Africa (e.g. Ashesi University, Makerere University, Dedan Kimathi University of Technology, AUST, AIST, Addis Ababa).

As of 24th October 2019, the Turing Institute announced that this work has been funded through a Turing Institute Senior AI Fellowship. This is the first Senior AI fellowship and it provides funding for five years.

The project partners are Element AI, Open ML, Professor Sylvie Delacroix and Data Science Africa.

Inclusive Project

There is no way that the team we’re building will be able to deliver on this agenda alone, so please join us in addressing these challenges!

Artificial vs Natural Systems [edit]

Let’s take a step back from artificial intelligence, and consider natural intelligence. Or even more generally, let’s consider the contrast between an artificial system and an natural system. The key difference between the two is that artificial systems are designed whereas natural systems are evolved.

Systems design is a major component of all Engineering disciplines. The details differ, but there is a single common theme: achieve your objective with the minimal use of resources to do the job. That provides efficiency. The engineering designer imagines a solution that requires the minimal set of components to achieve the result. A water pump has one route through the pump. That minimises the number of components needed. Redundancy is introduced only in safety critical systems, such as aircraft control systems. Students of biology, however, will be aware that in nature system-redundancy is everywhere. Redundancy leads to robustness. For an organism to survive in an evolving environment it must first be robust, then it can consider how to be efficient. Indeed, organisms that evolve to be too efficient at a particular task, like those that occupy a niche environment, are particularly vulnerable to extinction.

This notion is akin to the idea that only the best will survive, popularly encoded into an notion of evolution by Herbert Spencer’s quote.

Survival of the fittest

Herbet Spencer, 1864

Darwin himself never said “Survival of the Fittest” he talked about evolution by natural selection.

Non-survival of the non-fit

Evolution is better described as “non-survival of the non-fit”. You don’t have to be the fittest to survive, you just need to avoid the pitfalls of life. This is the first priority.

So it is with natural vs artificial intelligences. Any natural intelligence that was not robust to changes in its external environment would not survive, and therefore not reproduce. In contrast the artificial intelligences we produce are designed to be efficient at one specific task: control, computation, playing chess. They are fragile.

The first rule of a natural system is not be intelligent, it is “don’t be stupid”.

A mistake we make in the design of our systems is to equate fitness with the objective function, and to assume it is known and static. In practice, a real environment would have an evolving fitness function which would be unknown at any given time.

You can also read this blog post on Natural and Artificial Intelligence..

The first criterion of a natural intelligence is don’t fail, not because it has a will or intent of its own, but because if it had failed it wouldn’t have stood the test of time. It would no longer exist. In contrast, the mantra for artificial systems is to be more efficient. Our artificial systems are often given a single objective (in machine learning it is encoded in a mathematical function) and they aim to achieve that objective efficiently. These are different characteristics. Even if we wanted to incorporate don’t fail in some form, it is difficult to design for. To design for “don’t fail”, you have to consider every which way in which things can go wrong, if you miss one you fail. These cases are sometimes called corner cases. But in a real, uncontrolled environment, almost everything is a corner. It is difficult to imagine everything that can happen. This is why most of our automated systems operate in controlled environments, for example in a factory, or on a set of rails. Deploying automated systems in an uncontrolled environment requires a different approach to systems design. One that accounts for uncertainty in the environment and is robust to unforeseen circumstances.

Today’s Artificial Systems

The systems we produce today only work well when their tasks are pigeonholed, bounded in their scope. To achieve robust artificial intelligences we need new approaches to both the design of the individual components, and the combination of components within our AI systems. We need to deal with uncertainty and increase robustness. Today, it is easy to make a fool of an artificial intelligent agent, technology needs to address the challenge of the uncertain environment to achieve robust intelligences.

However, even if we find technological solutions for these challenges, it may be that the essence of human intelligence remains out of reach. It may be that the most quintessential element of our intelligence is defined by limitations. Limitations that computers have never experienced.

Claude Shannon developed the idea of information theory: the mathematics of information. He defined the amount of information we gain when we learn the result of a coin toss as a “bit” of information. A typical computer can communicate with another computer with a billion bits of information per second. Equivalent to a billion coin tosses per second. So how does this compare to us? Well, we can also estimate the amount of information in the English language. Shannon estimated that the average English word contains around 12 bits of information, twelve coin tosses, this means our verbal communication rates are only around the order of tens to hundreds of bits per second. Computers communicate tens of millions of times faster than us, in relative terms we are constrained to a bit of pocket money, while computers are corporate billionaires.

Our intelligence is not an island, it interacts, it infers the goals or intent of others, it predicts our own actions and how we will respond to others. We are social animals, and together we form a communal intelligence that characterises our species. For intelligence to be communal, our ideas to be shared somehow. We need to overcome this bandwidth limitation. The ability to share and collaborate, despite such constrained ability to communicate, characterises us. We must intellectually commune with one another. We cannot communicate all of what we saw, or the details of how we are about to react. Instead, we need a shared understanding. One that allows us to infer each other’s intent through context and a common sense of humanity. This characteristic is so strong that we anthropomorphise any object with which we interact. We apply moods to our cars, our cats, our environment. We seed the weather, volcanoes, trees with intent. Our desire to communicate renders us intellectually animist.

But our limited bandwidth doesn’t constrain us in our imaginations. Our consciousness, our sense of self, allows us to play out different scenarios. To internally observe how our self interacts with others. To learn from an internal simulation of the wider world. Empathy allows us to understand others’ likely responses without having the full detail of their mental state. We can infer their perspective. Self-awareness also allows us to understand our own likely future responses, to look forward in time, play out a scenario. Our brains contain a sense of self and a sense of others. Because our communication cannot be complete it is both contextual and cultural. When driving a car in the UK a flash of the lights at a junction concedes the right of way and invites another road user to proceed, whereas in Italy, the same flash asserts the right of way and warns another road user to remain.

Our main intelligence is our social intelligence, intelligence that is dedicated to overcoming our bandwidth limitation. We are individually complex, but as a society we rely on shared behaviours and oversimplification of our selves to remain coherent.

This nugget of our intelligence seems impossible for a computer to recreate directly, because it is a consequence of our evolutionary history. The computer, on the other hand, was born into a world of data, of high bandwidth communication. It was not there through the genesis of our minds and the cognitive compromises we made are lost to time. To be a truly human intelligence you need to have shared that journey with us.

Of course, none of this prevents us emulating those aspects of human intelligence that we observe in humans. We can form those emulations based on data. But even if an artificial intelligence can emulate humans to a high degree of accuracy it is a different type of intelligence. It is not constrained in the way human intelligence is. You may ask does it matter? Well, it is certainly important to us in many domains that there’s a human pulling the strings. Even in pure commerce it matters: the narrative story behind a product is often as important as the product itself. Handmade goods attract a price premium over factory made. Or alternatively in entertainment: people pay more to go to a live concert than for streaming music over the internet. People will also pay more to go to see a play in the theatre rather than a movie in the cinema.

In many respects I object to the use of the term Artificial Intelligence. It is poorly defined and means different things to different people. But there is one way in which the term is very accurate. The term artificial is appropriate in the same way we can describe a plastic plant as an artificial plant. It is often difficult to pick out from afar whether a plant is artificial or not. A plastic plant can fulfil many of the functions of a natural plant, and plastic plants are more convenient. But they can never replace natural plants.

In the same way, our natural intelligence is an evolved thing of beauty, a consequence of our limitations. Limitations which don’t apply to artificial intelligences and can only be emulated through artificial means. Our natural intelligence, just like our natural landscapes, should be treasured and can never be fully replaced.

Technical Consequence [edit]

Classical systems design assumes that the system is decomposable. That we can decompose the complex decision making process into distinct and independently designable parts. The composition of these parts gives us our final system.

Nicolas Negroponte, the original founder of MIT’s media lab used to write a column called ‘bits and atoms’. This referred to the ability of information to effect movement of goods in the physical world. It is this interaction where machine learning technologies have the possibility to bring most benefit.

Machine Learning Systems Design [edit]

The challenges of integrating different machine learning components into a whole that acts effectively as a system seem unresolved. In software engineering, separating parts of a system in this way is known as component-based software engineering. The core idea is that the different parts of the system can be independently designed according to a sub-specfication. This is sometimes known as separation of concerns. However, once the components are machine learning based, tighter coupling becomes a side effect of the learned nature of the system. For example if a driverless car’s detection of cyclist is dependent on its detection of the road surface, a change in the road surface detection algorithm will have downstream effects on the cyclist detection. Even if the road detection system has been improved by objective measures, the cyclist detection system may have become sensitive to the foibles of the previous version of road detection and will need to be retrained.

Most of our experience with deployment relies on some approximation to the component based model, this is also important for verification of the system. If the components of the system can be verified then the composed system can also, potentially, be verified.

The Data Crisis [edit]

Anecdotally, talking to data modelling scientists. Most say they spend 80% of their time acquiring and cleaning data. This is precipitating what I refer to as the “data crisis”. This is an analogy with software. The “software crisis” was the phenomenon of inability to deliver software solutions due to increasing complexity of implementation. There was no single shot solution for the software crisis, it involved better practice (scrum, test orientated development, sprints, code review), improved programming paradigms (object orientated, functional) and better tools (CVS, then SVN, then git).

However, these challenges aren’t new, they are merely taking a different form. From the computer’s perspective software is data. The first wave of the data crisis was known as the software crisis.

The Software Crisis

In the late sixties early software programmers made note of the increasing costs of software development and termed the challenges associated with it as the “Software Crisis”. Edsger Dijkstra referred to the crisis in his 1972 Turing Award winner’s address.

The major cause of the software crisis is that the machines have become several orders of magnitude more powerful! To put it quite bluntly: as long as there were no machines, programming was no problem at all; when we had a few weak computers, programming became a mild problem, and now we have gigantic computers, programming has become an equally gigantic problem.

Edsger Dijkstra (1930-2002), The Humble Programmer

The major cause of the data crisis is that machines have become more interconnected than ever before. Data access is therefore cheap, but data quality is often poor. What we need is cheap high-quality data. That implies that we develop processes for improving and verifying data quality that are efficient.

There would seem to be two ways for improving efficiency. Firstly, we should not duplicate work. Secondly, where possible we should automate work.

What I term “The Data Crisis” is the modern equivalent of this problem. The quantity of modern data, and the lack of attention paid to data as it is initially “laid down” and the costs of data cleaning are bringing about a crisis in data-driven decision making. This crisis is at the core of the challenge of technical debt in machine learning (Sculley et al. 2015).

Just as with software, the crisis is most correctly addressed by ‘scaling’ the manner in which we process our data. Duplication of work occurs because the value of data cleaning is not correctly recognised in management decision making processes. Automation of work is increasingly possible through techniques in “artificial intelligence”, but this will also require better management of the data science pipeline so that data about data science (meta-data science) can be correctly assimilated and processed. The Alan Turing institute has a program focussed on this area, AI for Data Analytics.

Computer Science Paradigm Shift

The next wave of machine learning is a paradigm shift in the way we think about computer science.

Classical computer science assumes that ‘data’ and ‘code’ are separate, and this is the foundation of secure computer systems. In machine learning, ‘data’ is ‘software’, so the decision making is directly influenced by the data. We are short-circuiting a fundamental assumption of computer science, we are breeching the code/data separation.

This means we need to revisit many of our assumptions and tooling around the machine learning process. In particular, we need new approaches to systems design, new approaches to programming languages that highlight the importance of data, and new approaches to systems security.

Peppercorns [edit]

Figure: A peppercorn is a system design failure which is not a bug, but a conformance to design specification that causes problems when the system is deployed in the real world with mischevious and adversarial actors.

Asking Siri “What is a trillion to the power of a thousand minus one?” leads to a 30 minute response1 consisting of only 9s. I found this out because my nine year old grabbed my phone and did it. The only way to stop Siri was to force closure. This is an interesting example of a system feature that’s not a bug, in fact it requires clever processing from Wolfram Alpha. But it’s an unexpected result from the system performing correctly.

This challenge of facing a circumstance that was unenvisaged in design but has consequences in deployment becomes far larger when the environment is uncontrolled. Or in the extreme case, where actions of the intelligent system effect the wider environment and change it.

These unforseen circumstances are likely to lead to need for much more efficient turn-around and update for our intelligent systems. Whether we are correcting for security flaws (which are bugs) or unenvisaged circumstantial challenges: an issue I’m referring to as peppercorns. Rapid deployment of system updates is required. For example, Apple have “fixed” the problem of Siri returning long numbers.

The challenge is particularly acute because of the scale at which we can deploy AI solutions. This means when something does go wrong, it may be going wrong in billions of households simultaneously.

You can also check this blog post on Decision Making and Diversity. and this blog post on Natural vs Artifical Intelligence..

The Three Ds of Machine Learning Systems Design [edit]

We can characterize the challenges for integrating machine learning within our systems as the three Ds. Decomposition, Data and Deployment.

You can also check my blog post on blog post on The 3Ds of Machine Learning Systems Design..

The first two components decomposition and data are interlinked, but we will first outline the decomposition challenge. Below we will mainly focus on supervised learning because this is arguably the technology that is best understood within machine learning.

Decomposition [edit]

Machine learning is not magical pixie dust, we cannot simply automate all decisions through data. We are constrained by our data (see below) and the models we use.2 Machine learning models are relatively simple function mappings that include characteristics such as smoothness. With some famous exceptions, e.g. speech and image data, inputs are constrained in the form of vectors and the model consists of a mathematically well-behaved function. This means that some careful thought has to be put in to the right sub-process to automate with machine learning. This is the challenge of decomposition of the machine learning system.

Any repetitive task is a candidate for automation, but many of the repetitive tasks we perform as humans are more complex than any individual algorithm can replace. The selection of which task to automate becomes critical and has downstream effects on our overall system design.

Pigeonholing

Figure: The machine learning systems decomposition process calls for separating a complex task into decomposable separate entities. A process we can think of as pigeonholing.

Some aspects to take into account are

  1. Can we refine the decision we need to a set of repetitive tasks where input information and output decision/value is well defined?
  2. Can we represent each sub-task we’ve defined with a mathematical mapping?

The representation necessary for the second aspect may involve massaging of the problem: feature selection or adaptation. It may also involve filtering out exception cases (perhaps through a pre-classification).

All else being equal, we’d like to keep our models simple and interpretable. If we can convert a complex mapping to a linear mapping through clever selection of sub-tasks and features this is a big win.

For example, Facebook have feature engineers, individuals whose main role is to design features they think might be useful for one of their tasks (e.g. newsfeed ranking, or ad matching). Facebook have a training/testing pipeline called FBLearner. Facebook have predefined the sub-tasks they are interested in, and they are tightly connected to their business model.

It is easier for Facebook to do this because their business model is heavily focused on user interaction. A challenge for companies that have a more diversified portfolio of activities driving their business is the identification of the most appropriate sub-task. A potential solution to feature and model selection is known as AutoML (Feurer et al., n.d.). Or we can think of it as using Machine Learning to assist Machine Learning. It’s also called meta-learning. Learning about learning. The input to the ML algorithm is a machine learning task, the output is a proposed model to solve the task.

One trap that is easy to fall in is too much emphasis on the type of model we have deployed rather than the appropriateness of the task decomposition we have chosen.

Recommendation: Conditioned on task decomposition, we should automate the process of model improvement. Model updates should not be discussed in management meetings, they should be deployed and updated as a matter of course. Further details below on model deployment, but model updating needs to be considered at design time. This is the domain of AutoML.

Figure: The answer to the question which comes first, the chicken or the egg is simple, they co-evolve (Popper 1963). Similarly, when we place components together in a complex machine learning system, they will tend to co-evolve and compensate for one another.

To form modern decision-making systems, many components are interlinked. We decompose our complex decision making into individual tasks, but the performance of each component is dependent on those upstream of it.

This naturally leads to co-evolution of systems; upstream errors can be compensated by downstream corrections.

To embrace this characteristic, end-to-end training could be considered. Why produce the best forecast by metrics when we can just produce the best forecast for our systems? End-to-end training can lead to improvements in performance, but it would also damage our systems decomposability and its interpretability, and perhaps its adaptability.

The less human interpretable our systems are, the harder they are to adapt to different circumstances or diagnose when there’s a challenge. The trade-off between interpretability and performance is a constant tension which we should always retain in our minds when performing our system design.

Combining Data and Systems Design [edit]

Data Science as Debugging [edit]

One challenge for existing information technology professionals is realizing the extent to which a software ecosystem based on data differs from a classical ecosystem. In particular, by ingesting data we bring unknowns/uncontrollables into our decision-making system. This presents opportunity for adversarial exploitation and unforeseen operation.

blog post on Data Science as Debugging.

Starting with the analysis of a data set, the nature of data science is somewhat difference from classical software engineering.

One analogy I find helpful for understanding the depth of change we need is the following. Imagine as a software engineer, you find a USB stick on the ground. And for some reason you know that on that USB stick is a particular API call that will enable you to make a significant positive difference on a business problem. You don’t know which of the many library functions on the USB stick are the ones that will help. And it could be that some of those library functions will hinder, perhaps because they are just inappropriate or perhaps because they have been placed there maliciously. The most secure thing to do would be to not introduce this code into your production system at all. But what if your manager told you to do so, how would you go about incorporating this code base?

The answer is very carefully. You would have to engage in a process more akin to debugging than regular software engineering. As you understood the code base, for your work to be reproducible, you should be documenting it, not just what you discovered, but how you discovered it. In the end, you typically find a single API call that is the one that most benefits your system. But more thought has been placed into this line of code than any line of code you have written before.

An enormous amount of debugging would be required. As the nature of the code base is understood, software tests to verify it also need to be constructed. At the end of all your work, the lines of software you write to actually interact with the software on the USB stick are likely to be minimal. But more thought would be put into those lines than perhaps any other lines of code in the system.

Even then, when your API code is introduced into your production system, it needs to be deployed in an environment that monitors it. We cannot rely on an individual’s decision making to ensure the quality of all our systems. We need to create an environment that includes quality controls, checks and bounds, tests, all designed to ensure that assumptions made about this foreign code base are remaining valid.

This situation is akin to what we are doing when we incorporate data in our production systems. When we are consuming data from others, we cannot assume that it has been produced in alignment with our goals for our own systems. Worst case, it may have been adversarially produced. A further challenge is that data is dynamic. So, in effect, the code on the USB stick is evolving over time.

It might see that this process is easy to formalize now, we simply need to check what the formal software engineering process is for debugging, because that is the current software engineering activity that data science is closest to. But when we look for a formalization of debugging, we find that there is none. Indeed, modern software engineering mainly focusses on ensuring that code is written without bugs in the first place.

Recommendation: Anecdotally, resolving a machine learning challenge requires 80% of the resource to be focused on the data and perhaps 20% to be focused on the model. But many companies are too keen to employ machine learning engineers who focus on the models, not the data. We should change our hiring priorities and training. Universities cannot provide the understanding of how to data-wrangle. Companies must fill this gap.

Figure: A reservoir of data has more value if the data is consumable. The data crisis can only be addressed if we focus on outputs rather than inputs.

Figure: For a data first architecture we need to clean our data at source, rather than individually cleaning data for each task. This involves a shift of focus from our inputs to our outputs. We should provide data streams that are consumable by many teams without purification.

Recommendation: We need to share best practice around data deployment across our teams. We should make best use of our processes where applicable, but we need to develop them to become data first organizations. Data needs to be cleaned at output not at input.

Deployment [edit]

Much of the academic machine learning systems point of view is based on a software systems point of view that is around 20 years out of date. In particular we build machine learning models on fixed training data sets, and we test them on stationary test data sets.

In practice modern software systems involve continuous deployment of models into an ever-evolving world of data. These changes are indicated in the software world by greater availability of technologies like streaming technologies.

Continuous Deployment

Once the decomposition is understood, the data is sourced and the models are created, the model code needs to be deployed.

To extend the USB stick analogy further, how would as software engineer deploy the code if they thought that the code might evolve in production? This is what data does. We cannot assume that the conditions under which we trained our model will be retained as we move forward, indeed the only constant we have is change.

This means that when any data dependent model is deployed into production, it requires continuous monitoring to ensure the assumptions of design have not been invalidated. Software changes are qualified through testing, in particular a regression test ensures that existing functionality is not broken by change. Since data is continually evolving, machine learning systems require ‘continual regression testing’: oversight by systems that ensure their existing functionality has not been broken as the world evolves around them. An approach we refer to as progression testing. Unfortunately, standards around ML model deployment yet been developed. The modern world of continuous deployment does rely on testing, but it does not recognize the continuous evolution of the world around us.

Progression tests are likely to be statistical tests in contrast to classical software tests. The tests should be monitoring model performance and quality measures. They could also monitor conformance to standardized fairness measures.

If the world has changed around our decision-making ecosystem, how are we alerted to those changes?

Recommendation: We establish best practice around model deployment. We need to shift our culture from standing up a software service, to standing up a data as a service. Data as a Service would involve continual monitoring of our deployed models in production. This would be regulated by ‘hypervisor’ systems3 that understand the context in which models are deployed and recognize when circumstances have changed, and models need retraining or restructuring.

Data Oriented Architectures [edit]

In a streaming architecture we shift from management of services, to management of data streams. Instead of worrying about availability of the services we shift to worrying about the quality of the data those services are producing.

Historically we’ve been software first, this is a necessary but insufficient condition for data first. We need to move from software-as-a-service to data-as-a-service, from service oriented architectures to data oriented architectures.

Streaming System

Characteristics of a streaming system include a move from pull updates to push updates, i.e. the computation is driven by a change in the input data rather than the service calling for input data when it decides to run a computation. Streaming systems operate on ‘rows’ of the data rather than ‘columns’. This is because the full column isn’t normally available as it changes over time. As an important design principle, the services themselves are stateless, they take their state from the streaming ecosystem. This ensures the inputs and outputs of given computations are easy to declare. As a result, persistence of the data is also handled by the streaming ecosystem and decisions around data retention or recomputation can be taken at the systems level rather than the component level.

Recommendation: We should consider a major re-architecting of systems around our services. In particular we should scope the use of a streaming architecture (such as Apache Kafka) that ensures data persistence and enables asynchronous operation of our systems.4 This would enable the provision of QC streams, and real time dash boards as well as hypervisors.

Importantly a streaming architecture implies the services we build are stateless, internal state is deployed on streams alongside external state. This allows for rapid assessment of other services’ data.

Apache Flink is a stream processing framework. Flink is a foundation for event driven processing. This gives a high throughput and low latency framework that operates on dataflows.

Data storage is handled by other systems such as Apache Kafka or AWS Kinesis.

stream.join(otherStream)
    .where(<KeySelector>)
    .equalTo(<KeySelector>)
    .window(<WindowAssigner>)
    .apply(<JoinFunction>)

Apache Flink allows operations on streams. For example, the join operation above. In a traditional data base management system, this join operation may be written in SQL and called on demand. In a streaming ecosystem, computations occur as and when the streams update.

The join is handled by the ecosystem surrounding the business logic.

Milan [edit]

Milan is a data-oriented programming language and runtime infrastructure.

https://github.com/amzn/milan

The Milan language is a DSL embedded in Scala. The output is an intermediate language that can be compiled to run on different target platforms. Currently there exists a single compiler that produces Flink applications.

The Milan runtime infrastructure compiles and runs Milan applications on a Flink cluster.

Trading System

As a simple example we’ll consider a high frequency trading system. Anne wishes to build a share trading system. She has access to a high frequency trading system which provides prices and allows trades at millisecond intervals. She wishes to build an automated trading system.

Let’s assume that price trading data is available as a data stream. But the price now is not the only information that Anne needs, she needs an estimate of the price in the future.

Figure: Anne has access to the share prices in the black stream but not in the blue stream. A hypothetical stream is the stream of future prices. Anne can define this hypothetical under constraints (latency, input etc). The need for a model is now exposed in the software infrastructure

Hypothetical Streams

We’ll call the future price a hypothetical stream.

A hypothetical stream is a desired stream of information which cannot be directly accessed. The lack of direct access may be because the events happen in the future, or there may be some latency between the event and the availability of the data.

Any hypothetical stream will only be provided as a prediction, ideally with an error bar.

The nature of the hypothetical Anne needs is dependent on her decision-making process. In Anne’s case it will depend over what period she is expecting her returns. In MDOP Anne specifies a hypothetical that is derived from the pricing stream.

It is not the price stream directly, but Anne looks for future predictions from the price stream, perhaps for price in T days’ time.

At this stage, this stream is merely typed as a hypothetical.

There are constraints on the hypothetical, they include: the input information, the upper limit of latency between input and prediction, and the decision Anne needs to make (how far ahead, what her upside, downside risks are). These three constraints mean that we can only recover an approximation to the hypothetical.

Hypothetical Advantage

What is the advantage to defining things in this way? By defining, clearly, the two streams as real and hypothetical variants of each other, we now enable automation of the deployment and any redeployment process. The hypothetical can be instantiated against the real, and design criteria can be constantly evaluated triggering retraining when necessary.

Let’s consider a ride sharing app, for example the SafeBoda system.

Anne is on her way home now; she wishes to hail a car using a ride sharing app.

The app is designed in the following way. On opening her app Anne is notified about drivers in the nearby neighborhood. She is given an estimate of the time a ride may take to come.

Given this information about driver availability, Anne may feel encouraged to enter a destination. Given this destination, a price estimate can be given. This price is conditioned on other riders that may wish to go in the same direction, but the price estimate needs to be made before the user agrees to the ride.

Business customer service constraints dictate that this price may not change after Anne’s order is confirmed.

In this simple system, several decisions are being made, each of them on the basis of a hypothetical.

When Anne calls for a ride, she is provided with an estimate based on the expected time a ride can be with her. But this estimate is made without knowing where Anne wants to go. There are constraints on drivers imposed by regional boundaries, reaching the end of their shift, or their current passengers mean that this estimate can only be a best guess.

This best guess may well be driven by previous data.

Ride Sharing: Service Oriented to Data Oriented [edit]

Figure: Service oriented architecture. The data access is buried in the cost allocation service. Data dependencies of the service cannot be found without trawling through the underlying code base.

The modern approach to software systems design is known as a service-oriented architectures (SOA). The idea is that software engineers are responsible for the availability and reliability of the API that accesses the service they own. Quality of service is maintained by rigorous standards around testing of software systems.

Figure: Data oriented architecture. Now the joins and the updates are exposed within the streaming ecosystem. We can programatically determine the factor graph which gives the thread through the model.

In data driven decision-making systems, the quality of decision-making is determined by the quality of the data. We need to extend the notion of service-oriented architecture to data-oriented architecture (DOA).

The focus in SOA is eliminating hard failures. Hard failures can occur due to bugs or systems overload. This notion needs to be extended in ML systems to capture soft failures associated with declining data quality, incorrect modeling assumptions and inappropriate re-deployments of models. We need to focus on data quality assessments. In data-oriented architectures engineering teams are responsible for the quality of their output data streams in addition to the availability of the service they support (Lawrence 2017). Quality here is not just accuracy, but fairness and explainability. This important cultural change would be capable of addressing both the challenge of technical debt (Sculley et al. 2015) and the social responsibility of ML systems.

Software development proceeds with a test-oriented culture. One where tests are written before software, and software is not incorporated in the wider system until all tests pass. We must apply the same standards of care to our ML systems, although for ML we need statistical tests for quality, fairness and consistency within the environment. Fortunately, the main burden of this testing need not fall to the engineers themselves: through leveraging classical statistics and emulation we will automate the creation and redeployment of these tests across the software ecosystem, we call this ML hypervision (WP5 ).

Modern AI can be based on ML models with many millions of parameters, trained on very large data sets. In ML, strong emphasis is placed on predictive accuracy whereas sister-fields such as statistics have a strong emphasis on interpretability. ML models are said to be ‘black boxes’ which make decisions that are not explainable.5

Figure: Data-oriented programing. There is a requirement for an estimate of the driver allocation to give a rough cost estimate before the user has confirmed the ride. In data-oriented programming, this is achieved through declaring a hypothetical stream which approximates the true driver allocation, but with restricted input information and constraints on the computational latency.

For the ride sharing system, we start to see a common issue with a more complex algorithmic decision-making system. Several decisions are being made multilple times. Let’s look at the decisions we need along with some design criteria.

  1. Driver Availability: Estimate time to arrival for Anne’s ride using Anne’s location and local available car locations. Latency 50 milliseconds
  2. Cost Estimate: Estimate cost for journey using Anne’s destination, location and local available car current destinations and availability. Latency 50 milliseconds
  3. Driver Allocation: Allocate car to minimize transport cost to destination. Latency 2 seconds.

So we need:

  1. a hypothetical to estimate availability. It is constrained by lacking destination information and a low latency requirement.
  2. a hypothetical to estimate cost. It is constrained by low latency requirement and

Simultaneously, drivers in this data ecosystem have an app which notifies them about new jobs and recommends them where to go.

Further advantages. Strategies for data retention (when to snapshot) can be set globally.

A few decisions need to be made in this system. First of all, when the user opens the app, the estimate of the time to the nearest ride may need to be computed quickly, to avoid latency in the service.

This may require a quick estimate of the ride availability.

Information Dynamics [edit]

With all the second guessing within a complex automated decision-making system, there are potential problems with information dynamics, the ‘closed loop’ problem, where the sub-systems are being approximated (second guessing) and predictions downstream are being affected.

This leads to the need for a closed loop analysis, for example, see the “Closed Loop Data Science” project led by Rod Murray-Smith at Glasgow.

The Machine Learning Bit [edit]

Emulation [edit]

Figure: Real world systems consiste of simulators, that capture our domain knowledge about how our systems operate. Different simulators run at different speeds and granularities.

In many real world systems, decisions are made through simulating the environment. Simulations may operate at different granularities. For example, simulations are used in weather forecasts and climate forecasts. The UK Met office uses the same code for both, but operates climate simulations one at greater spatial and temporal resolutions.

Figure: A statistical emulator is a system that reconstructs the simulation with a statistical model.

A statistical emulator is a data-driven model that learns about the underlying simulation. Importantly, learns with uncertainty, so it ‘knows what it doesn’t know’. In practice, we can call the emulator in place of the simulator. If the emulator ‘doesn’t know’, it can call the simulator for the answer.

Figure: A statistical emulator is a system that reconstructs the simulation with a statistical model. As well as reconstructing the simulation, a statistical emulator can be used to correlate with the real world.

Figure: In modern machine learning system design, the emulator may also consider the output of ML models (for monitoring bias or accuracy) and Operations Research models..

As well as reconstructing an individual simulator, the emulator can calibrate the simulation to the real world, by monitoring differences between the simulator and real data. This allows the emulator to characterise where the simulation can be relied on, i.e. we can validate the simulator.

Similarly, the emulator can adjudicate between simulations. This is known as multi-fidelity emulation. The emulator characterizes which emulations perform well where.

If all this modelling is done with judiscious handling of the uncertainty, the computational doubt, then the emulator can assist in desciding what experiment should be run next to aid a decision: should we run a simulator, in which case which one, or should we attempt to acquire data from a real world intervention.

Bayesian Inference by Rejection Sampling [edit]

One view of Bayesian inference is to assume we are given a mechanism for generating samples, where we assume that mechanism is representing on accurate view on the way we believe the world works.

This mechanism is known as our prior belief.

We combine our prior belief with our observations of the real world by discarding all those samples that are inconsistent with our prior. The likelihood defines mathematically what we mean by inconsistent with the prior. The higher the noise level in the likelihood, the looser the notion of consistent.

The samples that remain are considered to be samples from the posterior.

This approach to Bayesian inference is closely related to two sampling techniques known as rejection sampling and importance sampling. It is realized in practice in an approach known as approximate Bayesian computation (ABC) or likelihood-free inference.

In practice, the algorithm is often too slow to be practical, because most samples will be inconsistent with the data and as a result the mechanism has to be operated many times to obtain a few posterior samples.

However, in the Gaussian process case, when the likelihood also assumes Gaussian noise, we can operate this mechanism mathematically, and obtain the posterior density analytically. This is the benefit of Gaussian processes.

Figure: One view of Bayesian inference is we have a machine for generating samples (the prior), and we discard all samples inconsistent with our data, leaving the samples of interest (the posterior). The Gaussian process allows us to do this analytically.

Deep Emulation [edit]

Figure: A potential path of models in a machine learning system.

As a solution we can use of emulators. When constructing an ML system, software engineers, ML engineers, economists and operations researchers are explicitly defining relationships between variables of interest in the system. That implicitly defines a joint distribution, $p(\dataVector^*, \dataVector)$. In a decomposable system any sub-component may be defined as $p(\dataVector_\mathbf{i}|\dataVector_\mathbf{j})$ where $\dataVector_\mathbf{i}$ and $\dataVector_\mathbf{j}$ represent sub-sets of the full set of variables $\left\{\dataVector^*, \dataVector \right\}$. In those cases where the relationship is deterministic, the probability density would collapse to a vector-valued deterministic function, $\mappingFunctionVector_\mathbf{i}\left(\dataVector_\mathbf{j}\right)$.

Inter-variable relationships could be defined by, for example a neural network (machine learning), an integer program (operational research), or a simulation (supply chain). This makes probabilistic inference in this joint density for real world systems is either very hard or impossible.

Emulation is a form of meta-modelling: we construct a model of the model. We can define the joint density of an emulator as $s(\dataVector*, \dataVector)$, but if this probability density is to be an accurate representation of our system, it is likely to be prohibitively complex. Current practice is to design an emulator to deal with a specific question. This is done by fitting an ML model to a simulation from the the appropriate conditional distribution, $p(\dataVector_\mathbf{i}|\dataVector_\mathbf{j})$, which is intractable. The emulator provides an approximated answer of the form $s(\dataVector_\mathbf{i}|\dataVector_\mathbf{j})$. Critically, an emulator should incorporate its uncertainty about its approximation. So the emulator answer will be less certain than direct access to the conditional $p(\dataVector_i|\dataVector_j)$, but it may be sufficiently confident to act upon. Careful design of emulators to answer a given question leads to efficient diagnostics and understanding of the system. But in a complex interacting system an exponentially increasing number of questions can be asked. This calls for a system of automated construction of emulators which selects the right structure and redeploys the emulator as necessary. Rapid redeployment of emulators could exploit pre-existing emulators through transfer learning.

Automatically deploying these families of emulators for full system understanding is highly ambitious. It requires advances in engineering infrastructure, emulation and Bayesian optimization. However, the intermediate steps of developing this architecture also allow for automated monitoring of system accuracy and fairness. This facilitates AutoML on a component-wise basis which we can see as a simple implementation of AutoAI. The proposal is structured so that despite its technical ambition there is a smooth ramp of benefits to be derived across the programme of work.

In Applied Mathematics, the field studying these techniques is known as uncertainty quantification. The new challenge is the automation of emulator creation on demand to answer questions of interest and facilitate the system design, i.e. AutoAI through BSO.

At design stage, any particular AI task could be decomposed in multiple ways. Bayesian system optimization will assist both in determining the large-scale system design through exploring different decompositions and in refinement of the deployed system.

So far, most work on emulators has focussed on emulating a single component. Automated deployment and maintenance of ML systems requires networks of emulators that can be deployed and redeployed on demand depending on the particular question of interest. Therefore, the technical innovations we require are in the mathematical composition of emulator models (Damianou and Lawrence 2013; Perdikaris et al. 2017). Different chains of emulators will need to be rapidly composed to make predictions of downstream performance. This requires rapid retraining of emulators and propagation of uncertainty through the emulation pipeline a process we call deep emulation.

Recomposing the ML system requires structural learning of the network. By parameterizing covariance functions appropriately this can be done through Gaussian processes (e.g. (Damianou et al., n.d.)), but one could also consider Bayesian neural networks and other generative models, e.g. Generative Adversarial Networks (Goodfellow et al. 2014).

Figure: A potential path of models in a machine learning system.

Figure: A potential path of models in a machine learning system.

Figure: A statistical emulator is a system that reconstructs the simulation with a statistical model. As well as reconstructing the simulation, a statistical emulator can be used to correlate with the real world.

Conclusion

Today’s artificial intelligence is fundamentally Machine Learning Systems design, but the systems we are building will not fulfill the promises we are making for them. We are not yet ready to deploy automation in fully uncontrolled environments. Until we modify our approaches we will not be able to deliver on the promise. Until then, monitoring and upadate of deployed systems will be key to practical and safe AI.

References

Damianou, Andreas, Carl Henrik Ek, Michalis K. Titsias, and Neil D. Lawrence. n.d. “Manifold Relevance Determination.” In.

Damianou, Andreas, and Neil D. Lawrence. 2013. “Deep Gaussian Processes.” In, 31:207–15.

Feurer, Matthias, Aaron Klein, Katharina Eggensperger, Jost Tobias Springenberg, Manuel Blum, and Frank Hutter. n.d. “Efficient and Robust Automated Machine Learning.” In Advances in Neural Information Processing Systems.

Goodfellow, Ian, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. “Generative Adversarial Nets.” In Advances in Neural Information Processing Systems 27, edited by Z. Ghahramani, M. Welling, C. Cortes, N. D. Lawrence, and K. Q. Weinberger, 2672–80. Curran Associates, Inc.

Lawrence, Neil D. 2017. “Data Readiness Levels.” arXiv.

Perdikaris, Paris, Maziar Raissi, Andreas Damianou, Neil D. Lawrence, and George Em Karnidakis. 2017. “Nonlinear Information Fusion Algorithms for Data-Efficient Multi-Fidelity Modelling.” Proc. R. Soc. A 473 (20160751). https://doi.org/10.1098/rspa.2016.0751.

Popper, Karl R. 1963. Conjectures and Refutations: The Growth of Scientific Knowledge. London: Routledge.

Sculley, D., Gary Holt, Daniel Golovin, Eugene Davydov, Todd Phillips, Dietmar Ebner, Vinay Chaudhary, Michael Young, Jean-François Crespo, and Dan Dennison. 2015. “Hidden Technical Debt in Machine Learning Systems.” In Advances in Neural Information Processing Systems 28, edited by Corinna Cortes, Neil D. Lawrence, Daniel D. Lee, Masashi Sugiyama, and Roman Garnett, 2503–11. Curran Associates, Inc. http://papers.nips.cc/paper/5656-hidden-technical-debt-in-machine-learning-systems.pdf.


  1. Apple has fixed this issue so that Siri no longer does this.

  2. We can also become constrained by our tribal thinking, just as each of the other groups can.

  3. Emulation, or surrogate modelling, is one very promising approach to forming such a hypervisor. Emulators are models we fit to other models, often simulations, but the could also be other machine learning models. These models operate at the meta-level, not on the systems directly. This means they can be used to model how the sub-systems interact. As well as emulators we should consider real time dash boards, anomaly detection, mutlivariate analysis, data visualization and classical statistical approaches for hypervision of our deployed systems.

  4. These approaches are one area of focus for my own team’s research. A data first architecture is a prerequisite for efficient deployment of machine learning systems.

  5. See for example “The Dark Secret at the Heart of AI” in Technology Review.