%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 (2015) /EventType (Poster) /Description-Abstract (We propose rectified factor networks \050RFNs\051 to efficiently construct very sparse\054 non\055linear\054 high\055dimensional representations of the input\056 RFN models identify rare and small events\054 have a low interference between code units\054 have a small reconstruction error\054 and explain the data covariance structure\056 RFN learning is a generalized alternating minimization algorithm derived from the posterior regularization method which enforces non\055negative and normalized posterior means\056 We proof convergence and correctness of the RFN learning algorithm\056On benchmarks\054 RFNs are compared to other unsupervised methods like autoencoders\054 RBMs\054 factor analysis\054 ICA\054 and PCA\056 In contrast to previous sparse coding methods\054 RFNs yield sparser codes\054 capture the data\047s covariance structure more precisely\054 and have a significantly smaller reconstruction error\056 We test RFNs as pretraining technique of deep networks on different vision datasets\054 where RFNs were superior to RBMs and autoencoders\056 On gene expression data from two pharmaceutical drug discovery studies\054 RFNs detected small and rare gene modules that revealed highly relevant new biological insights which were so far missed by other unsupervised methods\056RFN package for GPU\057CPU is available at http\072\057\057www\056bioinf\056jku\056at\057software\057rfn\056) /Producer (PyPDF2) /Title (Rectified Factor Networks) /Date (2015) /ModDate (D\07220151218150419\05508\04700\047) /Published (2015) /Type (Conference Proceedings) /firstpage (1846) /Book (Advances in Neural Information Processing Systems 28) /Description (Paper accepted and presented at the Neural Information Processing Systems Conference \050http\072\057\057nips\056cc\057\051) /Editors (C\056 Cortes and N\056D\056 Lawrence and D\056D\056 Lee and M\056 Sugiyama and R\056 Garnett and R\056 Garnett) /Author (Djork\055Arn\351 Clevert\054 Andreas Mayr\054 Thomas Unterthiner\054 Sepp Hochreiter) /lastpage (1854) >> endobj 3 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 4 0 obj << /Contents 13 0 R /Parent 1 0 R /Resources 14 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 36 0 R 37 0 R 38 0 R 39 0 R 40 0 R 41 0 R 42 0 R 43 0 R 44 0 R 45 0 R ] /Type /Page >> endobj 5 0 obj << /Contents 46 0 R /Parent 1 0 R /Resources 47 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 72 0 R 73 0 R 74 0 R 75 0 R 76 0 R 77 0 R 78 0 R 79 0 R 80 0 R ] /Type /Page >> endobj 6 0 obj << /Contents 81 0 R /Parent 1 0 R /Resources 82 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 131 0 R 132 0 R 133 0 R 134 0 R 135 0 R 136 0 R 137 0 R 138 0 R 139 0 R 140 0 R 141 0 R 142 0 R 143 0 R 144 0 R 145 0 R 146 0 R 147 0 R 148 0 R 149 0 R 150 0 R ] /Type /Page >> endobj 7 0 obj << /Contents 151 0 R /Parent 1 0 R /Resources 152 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 189 0 R 190 0 R 191 0 R 192 0 R 193 0 R 194 0 R 195 0 R 196 0 R 197 0 R 198 0 R ] /Type /Page >> endobj 8 0 obj << /Contents 199 0 R /Parent 1 0 R /Resources 200 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 201 0 R 202 0 R 203 0 R 204 0 R 205 0 R 206 0 R 207 0 R 208 0 R 209 0 R 210 0 R 211 0 R 212 0 R 213 0 R 214 0 R 215 0 R ] /Type /Page >> endobj 9 0 obj << /Contents 216 0 R /Parent 1 0 R /Resources 217 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 218 0 R 219 0 R 220 0 R ] /Type /Page >> endobj 10 0 obj << /Contents 221 0 R /Parent 1 0 R /Resources 222 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 323 0 R 324 0 R 325 0 R 326 0 R 327 0 R 328 0 R 329 0 R 330 0 R 331 0 R 332 0 R 333 0 R 334 0 R 335 0 R ] /Type /Page >> endobj 11 0 obj << /Contents 336 0 R /Parent 1 0 R /Resources 337 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 387 0 R 388 0 R 389 0 R 390 0 R 391 0 R 392 0 R ] /Type /Page >> endobj 12 0 obj << /Contents 393 0 R /Parent 1 0 R /Type /Page /Resources 394 0 R /MediaBox [ 0 0 612 792 ] >> endobj 13 0 obj << /Length 3192 /Filter /FlateDecode >> stream xڍYI6mZ4wsgĞL~sp|`S7E*\ijў9* Z`nϛc0gaexVo,U:kls7/^']Q&, rG, gGwS z{=R(w״7?_~dQ>~ fk ~~ljDϳ(#8(eaR3v\ܵ0jxd~cc?(KXQFUZ`vH`L ⹝瑧]wu;[VE{s8?r'-I"0>9"L -B%E7u~Uf#KzӴew+]QFe*:-AQMy._uz7t}k0LN8J9NX<4:ݑ.MemVnmVqeHB4b.S? EQɲzDW*DJS*`ZޡmMמ92,, O):`̆.},貁]g*TEv-}p~_uS/*\AΓ.?NrogM٦.*@qhM=ab؍9ևa o֦v v<%M[&_h틪9kw??cv< tjՓ[;ܘԥR=>Q%dP۾FQEY''0ӶS<CUXu* Rx,=e5>fʷ0!s;m"n+~e%x%ոy`ok_=lM{_ͦm-~Yr8 #&ۡAݖߛ~eN"aIqka0@l-A&Dͤp"Z´MQwt 8" fLqtbk+g5]r&N49:FN"r)L.wLmNPs4)=_8,3b;G$*SCߘzR{R- d˷2 Ǹ8=Ro):7hWwо}#w|9hD⋷ ocgc$@Em";Vt8G]rDAH9iunŵ#3[riyf8_\ObE2tv[sn*I&qJ2nLN /3I/{@Y#$=Yd zSj`s>ژ.2% }rb8;N툆;WunF8>{=pb]뒵G0o~Q@^VF1JΏ;DYe;# VNx% Ǝ:-wd $GWSZSOy82Fy]Ykb+n-y%L1~LO2E`vNdKR]$j3 i]8+ݥ.$fNWʊp':h2`yB.n.XT gђST
=*2:Б^s1w1gKo%uh6YT
S3R # {hM(е4cc,ZnB1U@
d6s2!pbkfRTRviJ:L e-V S49W>էIv<>ZqGz`ZkGRCCRњ3?3wJ=4@c Hl86HSWJ6$-|!hJNߛRHK6vh苙Hbigah}da#:S%PH+&g@
ück;vxVpgpݙtӋew1Z35v2^ !7㺘Q)#燍\RѺ:;U`ǮTCy-U2ngΔYurJL+ w;(#eD60uTpvmI:])ף{is0plԦ%a|R_"D^IӝV6<]z3\y-1?ơ{'s5B_#"I"e(#ⵛQPمK#oKc5%t_Xz#"Kq1 qHD(SٵB咑4<3rqJ eݦ>uEdT %xksf8p(u/f{©q<Шb1;*FC~kj!8U\Y~'-lѻSIːi}JW'lCt?93u,A9\uO6LᔚNPzn/}юb`?WKMcKϟ_(֫X&G :Pۤ{@Dw^OV+9R4AKq@cۋG&c74t絪DhΘջڵ-mqv5}'+
C ^`VE