What conditions on a training set ensure that there is a unique optimal linear hypothesis map for linear regression?
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Linear regression uses the squared error loss to measure the quality of a linear hypothesis map. We learn the weights [math]\weights[/math] of a linear map via ERM using a training set [math]\dataset[/math] that consists of [math]\samplesize=100[/math] data points. Each data point is characterized by [math]\featurelen=5[/math] features and a numeric label.
Is there a unique choice for the weights [math]\weights[/math] that results in a linear predictor with minimum average squared error loss on the training set [math]\dataset[/math])?
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Consider a training set of [math]\samplesize[/math] datapoints, each characterized by a single numeric feature [math]\feature[/math] and numeric label [math]\truelabel[/math]. We learn hypothesis map of the form [math]h(\feature) = \feature + b[/math] with some bias [math]b \in \mathbb{R}[/math].
Can you write down a formula for the optimal [math]b[/math], that minimizes the average squared error on training data [math]\big(\feature^{(1)},\truelabel^{(1)} \big),\ldots,\big(\feature^{(\samplesize)},\truelabel^{(\samplesize)}\big)[/math].
[math] % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Consider polynomial regression for data points with a single numeric feature [math]\feature \in \mathbb{R}[/math] and numeric label [math]\truelabel[/math]. Here, polynomial regression is equivalent to linear regression using the transformed feature vectors [math]\featurevec = \big(\feature^{0},\feature^{1},\ldots,\feature^{\featuredim-1}\big)^{T}[/math].
Given a dataset [math] \dataset= \big(\feature^{(1)},\truelabel^{(1)}\big),\ldots,\big(\feature^{(\samplesize)},\truelabel^{(\samplesize)}\big)[/math], we construct the feature matrix [math]\featuremtx =\big(\featurevec^{(1)},\ldots,\featurevec^{(\samplesize)}\big) \in \mathbb{R}^{\samplesize \times \samplesize}[/math] with its [math]\sampleidx[/math]th column given by the feature vector [math]\featurevec^{(\sampleidx)}[/math].
Verify that this feature matrix is a Vandermonde matrix [1]?
How is the determinant of the feature matrix related to the features and labels of data points in the dataset [math]\dataset[/math]?
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Consider a training set that consists of data points [math]\big(\feature^{(\sampleidx)},\truelabel^{(\sampleidx)} \big)[/math], for [math]\sampleidx = 1,\ldots,\samplesize=100[/math], that are obtained as realizations of iid RVs. The common probability distribution of these RVs is defined by a random data point [math](\feature,\truelabel)[/math]. The feature [math]\feature[/math] of this random data point is a standard Gaussian RV with zero mean and unit variance. The label of a data point is modelled as [math]\truelabel = \feature + e[/math] with Gaussian noise [math]e \sim \mathcal{N}(0,1)[/math]. The feature [math]\feature[/math] and noise [math]e[/math] are statistically independent.
We evaluate the specific hypothesis [math]h(\feature)=0[/math] (which outputs [math]0[/math] no matter what the feature value [math]\feature[/math] is) bythe training error [math]\trainerror = (1/\samplesize) \sum_{\sampleidx=1}^{\samplesize} \big( \truelabel^{(\sampleidx)} - h \big( \feature^{(\sampleidx)} \big) \big)^2[/math]. Note that [math]\trainerror[/math] is the average squared error loss incurred by hypothesis [math]h[/math] on the datapoints [math]\big(\feature^{(\sampleidx)},\truelabel^{(\sampleidx)} \big)[/math], for [math]\sampleidx = 1,\ldots,\samplesize=100[/math].
What is the probability that the training error [math]\trainerror[/math] is at least [math]20[/math] than the expected (squared error) loss [math]\expect \big\{ \big( \truelabel - h(\feature) \big)^{2} \big \}[/math]?
What is the mean (expected value) and variance of the training error ?
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Let us consider a fictional (idel) optimization method that can be represented as a filter [math]\mathcal{F}[/math]. This filter [math]\mathcal{F}[/math] reads in a real-valued objective function [math]f(\cdot)[/math], defined for all parameter vectors [math]\vw\in\mathbb{R}^{\featuredim}[/math]. The output of the filter [math]\mathcal{F}[/math] is another real-valued function [math]\hat{f}(\vw)[/math] that is defined point-wise as
Verify that the filter [math]\mathcal{F}[/math] is shift or translation invariant, i.e., [math]\mathcal{F}[/math] commutes with a translation [math]f'(\weights) \defeq f(\weights + \weights^{(o)})[/math] with an arbitrary but fixed (reference) vector [math]\weights^{(o)} \in \mathbb{R}^{\featuredim}[/math].
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Consider a linear regression method that uses ERM to learn weights [math]\widehat{\weights}[/math] of a linear hypothesis map [math]h(\featurevec) =\weights^{T} \featurevec[/math]. The weights are learnt by minimizing the average squared error loss incurred by [math]h[/math] on a training set that is constituted by the data points [math]\big( \featurevec^{(\sampleidx)}, \truelabel^{(\sampleidx)} \big)[/math] for [math]\sampleidx=1,\ldots, 100[/math]. Someimtes it is useful to assign sample-weights [math]\sampleweight{\sampleidx}[/math] to the data points and learn [math]\widehat{\weights}[/math]. These sample-weights reflect varying levels of importance or relevance of different data points.
For simplicity we use the sample weights [math]\sampleweight{\sampleidx} = 2 \alpha \in [0,1][/math] for [math]\sampleidx=1,\ldots,50[/math] and [math]\sampleweight{\sampleidx} = 2(1 - \alpha)[/math] for [math]\sampleidx=51,\ldots,100[/math].
Can you find a closed-form expression (similar to equ_close_form_lin_reg) for the weights [math]\widehat{\weights}^{(\alpha)}[/math] that minimize the weighted average squared error
for different [math]\alpha[/math]?
[math] \( % Generic syms \newcommand\defeq{:=} \newcommand{\Tt}[0]{\boldsymbol{\theta}} \newcommand{\XX}[0]{{\cal X}} \newcommand{\ZZ}[0]{{\cal Z}} \newcommand{\vx}[0]{{\bf x}} \newcommand{\vv}[0]{{\bf v}} \newcommand{\vu}[0]{{\bf u}} \newcommand{\vs}[0]{{\bf s}} \newcommand{\vm}[0]{{\bf m}} \newcommand{\vq}[0]{{\bf q}} \newcommand{\mX}[0]{{\bf X}} \newcommand{\mC}[0]{{\bf C}} \newcommand{\mA}[0]{{\bf A}} \newcommand{\mL}[0]{{\bf L}} \newcommand{\fscore}[0]{F_{1}} \newcommand{\sparsity}{s} \newcommand{\mW}[0]{{\bf W}} \newcommand{\mD}[0]{{\bf D}} \newcommand{\mZ}[0]{{\bf Z}} \newcommand{\vw}[0]{{\bf w}} \newcommand{\D}[0]{{\mathcal{D}}} \newcommand{\mP}{\mathbf{P}} \newcommand{\mQ}{\mathbf{Q}} \newcommand{\E}[0]{{\mathbb{E}}} \newcommand{\vy}[0]{{\bf y}} \newcommand{\va}[0]{{\bf a}} \newcommand{\vn}[0]{{\bf n}} \newcommand{\vb}[0]{{\bf b}} \newcommand{\vr}[0]{{\bf r}} \newcommand{\vz}[0]{{\bf z}} \newcommand{\N}[0]{{\mathcal{N}}} \newcommand{\vc}[0]{{\bf c}} \newcommand{\bm}{\boldsymbol} % Statistics and Probability Theory \newcommand{\errprob}{p_{\rm err}} \newcommand{\prob}[1]{p({#1})} \newcommand{\pdf}[1]{p({#1})} \def \expect {\mathbb{E} } % Machine Learning Symbols \newcommand{\biasterm}{B} \newcommand{\varianceterm}{V} \newcommand{\neighbourhood}[1]{\mathcal{N}^{(#1)}} \newcommand{\nrfolds}{k} \newcommand{\mseesterr}{E_{\rm est}} \newcommand{\bootstrapidx}{b} %\newcommand{\modeldim}{r} \newcommand{\modelidx}{l} \newcommand{\nrbootstraps}{B} \newcommand{\sampleweight}[1]{q^{(#1)}} \newcommand{\nrcategories}{K} \newcommand{\splitratio}[0]{{\rho}} \newcommand{\norm}[1]{\Vert {#1} \Vert} \newcommand{\sqeuclnorm}[1]{\big\Vert {#1} \big\Vert^{2}_{2}} \newcommand{\bmx}[0]{\begin{bmatrix}} \newcommand{\emx}[0]{\end{bmatrix}} \newcommand{\T}[0]{\text{T}} \DeclareMathOperator*{\rank}{rank} %\newcommand\defeq{:=} \newcommand\eigvecS{\hat{\mathbf{u}}} \newcommand\eigvecCov{\mathbf{u}} \newcommand\eigvecCoventry{u} \newcommand{\featuredim}{n} \newcommand{\featurelenraw}{\featuredim'} \newcommand{\featurelen}{\featuredim} \newcommand{\samplingset}{\mathcal{M}} \newcommand{\samplesize}{m} \newcommand{\sampleidx}{i} \newcommand{\nractions}{A} \newcommand{\datapoint}{\vz} \newcommand{\actionidx}{a} \newcommand{\clusteridx}{c} \newcommand{\sizehypospace}{D} \newcommand{\nrcluster}{k} \newcommand{\nrseeds}{s} \newcommand{\featureidx}{j} \newcommand{\clustermean}{{\bm \mu}} \newcommand{\clustercov}{{\bm \Sigma}} \newcommand{\target}{y} \newcommand{\error}{E} \newcommand{\augidx}{b} \newcommand{\task}{\mathcal{T}} \newcommand{\nrtasks}{T} \newcommand{\taskidx}{t} \newcommand\truelabel{y} \newcommand{\polydegree}{r} \newcommand\labelvec{\vy} \newcommand\featurevec{\vx} \newcommand\feature{x} \newcommand\predictedlabel{\hat{y}} \newcommand\dataset{\mathcal{D}} \newcommand\trainset{\dataset^{(\rm train)}} \newcommand\valset{\dataset^{(\rm val)}} \newcommand\realcoorspace[1]{\mathbb{R}^{\text{#1}}} \newcommand\effdim[1]{d_{\rm eff} \left( #1 \right)} \newcommand{\inspace}{\mathcal{X}} \newcommand{\sigmoid}{\sigma} \newcommand{\outspace}{\mathcal{Y}} \newcommand{\hypospace}{\mathcal{H}} \newcommand{\emperror}{\widehat{L}} \newcommand\risk[1]{\expect \big \{ \loss{(\featurevec,\truelabel)}{#1} \big\}} \newcommand{\featurespace}{\mathcal{X}} \newcommand{\labelspace}{\mathcal{Y}} \newcommand{\rawfeaturevec}{\mathbf{z}} \newcommand{\rawfeature}{z} \newcommand{\condent}{H} \newcommand{\explanation}{e} \newcommand{\explainset}{\mathcal{E}} \newcommand{\user}{u} \newcommand{\actfun}{\sigma} \newcommand{\noisygrad}{g} \newcommand{\reconstrmap}{r} \newcommand{\predictor}{h} \newcommand{\eigval}[1]{\lambda_{#1}} \newcommand{\regparam}{\lambda} \newcommand{\lrate}{\alpha} \newcommand{\edges}{\mathcal{E}} \newcommand{\generror}{E} \DeclareMathOperator{\supp}{supp} %\newcommand{\loss}[3]{L({#1},{#2},{#3})} \newcommand{\loss}[2]{L\big({#1},{#2}\big)} \newcommand{\clusterspread}[2]{L^{2}_{\clusteridx}\big({#1},{#2}\big)} \newcommand{\determinant}[1]{{\rm det}({#1})} \DeclareMathOperator*{\argmax}{argmax} \DeclareMathOperator*{\argmin}{argmin} \newcommand{\itercntr}{r} \newcommand{\state}{s} \newcommand{\statespace}{\mathcal{S}} \newcommand{\timeidx}{t} \newcommand{\optpolicy}{\pi_{*}} \newcommand{\appoptpolicy}{\hat{\pi}} \newcommand{\dummyidx}{j} \newcommand{\gridsizex}{K} \newcommand{\gridsizey}{L} \newcommand{\localdataset}{\mathcal{X}} \newcommand{\reward}{r} \newcommand{\cumreward}{G} \newcommand{\return}{\cumreward} \newcommand{\action}{a} \newcommand\actionset{\mathcal{A}} \newcommand{\obstacles}{\mathcal{B}} \newcommand{\valuefunc}[1]{v_{#1}} \newcommand{\gridcell}[2]{\langle #1, #2 \rangle} \newcommand{\pair}[2]{\langle #1, #2 \rangle} \newcommand{\mdp}[5]{\langle #1, #2, #3, #4, #5 \rangle} \newcommand{\actionvalue}[1]{q_{#1}} \newcommand{\transition}{\mathcal{T}} \newcommand{\policy}{\pi} \newcommand{\charger}{c} \newcommand{\itervar}{k} \newcommand{\discount}{\gamma} \newcommand{\rumba}{Rumba} \newcommand{\actionnorth}{\rm N} \newcommand{\actionsouth}{\rm S} \newcommand{\actioneast}{\rm E} \newcommand{\actionwest}{\rm W} \newcommand{\chargingstations}{\mathcal{C}} \newcommand{\basisfunc}{\phi} \newcommand{\augparam}{B} \newcommand{\valerror}{E_{v}} \newcommand{\trainerror}{E_{t}} \newcommand{\foldidx}{b} \newcommand{\testset}{\dataset^{(\rm test)} } \newcommand{\testerror}{E^{(\rm test)}} \newcommand{\nrmodels}{M} \newcommand{\benchmarkerror}{E^{(\rm ref)}} \newcommand{\lossfun}{L} \newcommand{\datacluster}[1]{\mathcal{C}^{(#1)}} \newcommand{\cluster}{\mathcal{C}} \newcommand{\bayeshypothesis}{h^{*}} \newcommand{\featuremtx}{\mX} \newcommand{\weight}{w} \newcommand{\weights}{\vw} \newcommand{\regularizer}{\mathcal{R}} \newcommand{\decreg}[1]{\mathcal{R}_{#1}} \newcommand{\naturalnumbers}{\mathbb{N}} \newcommand{\featuremapvec}{{\bf \Phi}} \newcommand{\featuremap}{\phi} \newcommand{\batchsize}{B} \newcommand{\batch}{\mathcal{B}} \newcommand{\foldsize}{B} \newcommand{\nriter}{R} [/math]
Consider data points characterized by single numeric feature [math]\feature[/math] and label [math]\truelabel[/math]. We learn a hypothesis map of the form [math]h(\feature) = \feature + b[/math] with some bias [math]b \in \mathbb{R}[/math].
Can you write down a formula for the optimal [math]b[/math], that minimizes the average absolute error on training data [math]\big(\feature^{(1)},\truelabel^{(1)} \big),\ldots,\big(\feature^{(\samplesize)},\truelabel^{(\samplesize)}\big)[/math].