%PDF-1.3 1 0 obj << /Kids [ 4 0 R 5 0 R 6 0 R 7 0 R 8 0 R 9 0 R 10 0 R 11 0 R 12 0 R ] /Type /Pages /Count 9 >> endobj 2 0 obj << /Subject (Neural Information Processing Systems http\072\057\057nips\056cc\057) /Publisher (Curran Associates\054 Inc\056) /Language (en\055US) /Created (2017) /EventType (Poster) /Description-Abstract (This paper studies empirical risk minimization \050ERM\051 problems for large\055scale datasets and incorporates the idea of adaptive sample size methods to improve the guaranteed convergence bounds for first\055order stochastic and deterministic methods\056 In contrast to traditional methods that attempt to solve the ERM problem corresponding to the full dataset directly\054 adaptive sample size schemes start with a small number of samples and solve the corresponding ERM problem to its statistical accuracy\056 The sample size is then grown geometrically \055\055 e\056g\056\054 scaling by a factor of two \055\055 and use the solution of the previous ERM as a warm start for the new ERM\056 Theoretical analyses show that the use of adaptive sample size methods reduces the overall computational cost of achieving the statistical accuracy of the whole dataset for a broad range of deterministic and stochastic first\055order methods\056 The gains are specific to the choice of method\056 When particularized to\054 e\056g\056\054 accelerated gradient descent and stochastic variance reduce gradient\054 the computational cost advantage is a logarithm of the number of training samples\056 Numerical experiments on various datasets confirm theoretical claims and showcase the gains of using the proposed adaptive sample size scheme\056) /Producer (PyPDF2) /Title (First\055Order Adaptive Sample Size Methods to Reduce Complexity of Empirical Risk Minimization) /Date (2017) /ModDate (D\07220180213005813\05508\04700\047) /Published (2017) /Type (Conference Proceedings) /firstpage (2060) /Book (Advances in Neural Information Processing Systems 30) /Description (Paper accepted and presented at the Neural Information Processing Systems Conference \050http\072\057\057nips\056cc\057\051) /Editors (I\056 Guyon and U\056V\056 Luxburg and S\056 Bengio and H\056 Wallach and R\056 Fergus and S\056 Vishwanathan and R\056 Garnett) /Author (Aryan Mokhtari\054 Alejandro Ribeiro) /lastpage (2068) >> endobj 3 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 4 0 obj << /Parent 1 0 R /Contents 13 0 R /Type /Page /Resources 14 0 R /MediaBox [ 0 0 612 792 ] >> endobj 5 0 obj << /Parent 1 0 R /Contents 59 0 R /Type /Page /Resources 60 0 R /MediaBox [ 0 0 612 792 ] >> endobj 6 0 obj << /Parent 1 0 R /Contents 124 0 R /Type /Page /Resources 125 0 R /MediaBox [ 0 0 612 792 ] >> endobj 7 0 obj << /Parent 1 0 R /Contents 138 0 R /Type /Page /Resources 139 0 R /MediaBox [ 0 0 612 792 ] >> endobj 8 0 obj << /Parent 1 0 R /Contents 146 0 R /Type /Page /Resources 147 0 R /MediaBox [ 0 0 612 792 ] >> endobj 9 0 obj << /Parent 1 0 R /Contents 148 0 R /Type /Page /Resources 149 0 R /MediaBox [ 0 0 612 792 ] >> endobj 10 0 obj << /Parent 1 0 R /Contents 150 0 R /Type /Page /Resources 151 0 R /MediaBox [ 0 0 612 792 ] >> endobj 11 0 obj << /Parent 1 0 R /Contents 152 0 R /Type /Page /Resources 153 0 R /MediaBox [ 0 0 612 792 ] >> endobj 12 0 obj << /Parent 1 0 R /Contents 244 0 R /Type /Page /Resources 245 0 R /MediaBox [ 0 0 612 792 ] >> endobj 13 0 obj << /Length 4738 /Filter /FlateDecode >> stream x[rF+0sb,F^vK>@ZM4 Ny/ IQlnܗүn E<#7KcVO ͂ÅE{7̰yaai"vn7C7toMvC^"pr8Tv\oW]3y¾p߹7ry5\H#7]~v O?#ne{[wW^}lF >J/m~er-kDܕ{ H`? àp Z>}ڏ%%eUS_40
[Ip(J}7wJc~ay6Gyq܊Ix]{a(]UB);뷯..mܯ }C"SҰ젲P@83k%Fmx(z1<1 ,v۴e/7b`?E釪²Tm7o``7X?㮵~a
ZY)YwOl=lM_UreKWz6jJm`+)oNv}|*^kf=T}Rdހ#%.i/9~z4w_47&D?Դd2D:`{9+?ր;aTqpux{,C3lnl#ϭ
*/~RFYPղ_'X/|F mbhXOZUE~Uv}sF9`H1rU-PKlv{=KPZ v2C69MVsNV;4
oȯZm?<[őc^D$!_.w0*^]>)y7X7Nl,P@,w~u0U
_!
G_儧 Sd.'5|Uٰr5B3Wb#"[Z:-VS]ZcN,5G(ttVaf&-:.B ؠTk o+vTIX"(5
|