%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 (2016) /EventType (Poster) /Description-Abstract (Large labeled training sets are the critical building blocks of supervised learning methods and are key enablers of deep learning techniques\056 For some applications\054 creating labeled training sets is the most time\055consuming and expensive part of applying machine learning\056 We therefore propose a paradigm for the programmatic creation of training sets called data programming in which users provide a set of labeling functions\054 which are programs that heuristically label subsets of the data\054 but that are noisy and may conflict\056 By viewing these labeling functions as implicitly describing a generative model for this noise\054 we show that we can recover the parameters of this model to \042denoise\042 the generated training set\054 and establish theoretically that we can recover the parameters of these generative models in a handful of settings\056 We then show how to modify a discriminative loss function to make it noise\055aware\054 and demonstrate our method over a range of discriminative models including logistic regression and LSTMs\056 Experimentally\054 on the 2014 TAC\055KBP Slot Filling challenge\054 we show that data programming would have led to a new winning score\054 and also show that applying data programming to an LSTM model leads to a TAC\055KBP score almost 6 F1 points over a state\055of\055the\055art LSTM baseline \050and into second place in the competition\051\056 Additionally\054 in initial user studies we observed that data programming may be an easier way for non\055experts to create machine learning models when training data is limited or unavailable\056) /Producer (PyPDF2) /Title (Data Programming\072 Creating Large Training Sets\054 Quickly) /Date (2016) /ModDate (D\07220170112153238\05508\04700\047) /Published (2016) /Type (Conference Proceedings) /firstpage (3567) /Book (Advances in Neural Information Processing Systems 29) /Description (Paper accepted and presented at the Neural Information Processing Systems Conference \050http\072\057\057nips\056cc\057\051) /Editors (D\056D\056 Lee and M\056 Sugiyama and U\056V\056 Luxburg and I\056 Guyon and R\056 Garnett) /Author (Alexander J\056 Ratner\054 Christopher M\056 De Sa\054 Sen Wu\054 Daniel Selsam\054 Christopher R\351) /lastpage (3575) >> 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 [ 39 0 R 40 0 R 41 0 R ] /Type /Page >> endobj 5 0 obj << /Contents 42 0 R /Parent 1 0 R /Resources 43 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 44 0 R 45 0 R 46 0 R 47 0 R 48 0 R 49 0 R 50 0 R 51 0 R 52 0 R 53 0 R 54 0 R 55 0 R 56 0 R 57 0 R 58 0 R 59 0 R 60 0 R 61 0 R ] /Type /Page >> endobj 6 0 obj << /Contents 62 0 R /Parent 1 0 R /Resources 63 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 76 0 R 77 0 R 78 0 R 79 0 R 80 0 R 81 0 R 82 0 R 83 0 R 84 0 R 85 0 R 86 0 R 87 0 R 88 0 R 89 0 R 90 0 R 91 0 R 92 0 R 93 0 R ] /Type /Page >> endobj 7 0 obj << /Contents 94 0 R /Parent 1 0 R /Resources 95 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 100 0 R 101 0 R 102 0 R 103 0 R 104 0 R 105 0 R 106 0 R ] /Type /Page >> endobj 8 0 obj << /Contents 107 0 R /Parent 1 0 R /Resources 108 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 109 0 R 110 0 R 111 0 R 112 0 R 113 0 R 114 0 R 115 0 R 116 0 R 117 0 R 118 0 R ] /Type /Page >> endobj 9 0 obj << /Contents 119 0 R /Parent 1 0 R /Resources 120 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 125 0 R 126 0 R 127 0 R 128 0 R 129 0 R 130 0 R 131 0 R 132 0 R ] /Type /Page >> endobj 10 0 obj << /Contents 133 0 R /Parent 1 0 R /Resources 134 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 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 ] /Type /Page >> endobj 11 0 obj << /Contents 144 0 R /Parent 1 0 R /Resources 145 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 146 0 R 147 0 R 148 0 R 149 0 R 150 0 R 151 0 R ] /Type /Page >> endobj 12 0 obj << /Contents 152 0 R /Parent 1 0 R /Type /Page /Resources 153 0 R /MediaBox [ 0 0 612 792 ] >> endobj 13 0 obj << /Length 3345 /Filter /FlateDecode >> stream xڕZK [U(ɧulT^8{Z^=zx@z& .MtMh(ۥQfq+7H0ϒ]ow7?8Jwq8QNH tQ rk (Y=io: 3v8sin5# E>UAovpgᆚ?OU~FoؗyJ֯E0=Qjww]p;b"Iw:Ԃ=I[h"`>0v ox[js?S1mek?L_$-qW0A('qFhH8 l?T3I!a w[)J;OM9?xЖˈy'd,Twq0)Fwba$ؔ%`*Ow ps,̓-j@nzN<諱*LM}SU<(uW|hlw$0mh5Xښޭ_eԕ7H<=ۚzNx֞58AqYu=Ԝ5{ r< E?,Z:+uC mh40B+9 X|j扤m_m;^M?R4.T>Y;2hLqZK8(4{sߝ;@ALY=6D<7,79FY~S|kv5dDE#bLOwOۋi*)#sW*Qǯ:L0/V,ө*NԜg'k>-믚Ӌp@!tϲkLNvBZ<ѕqxrq]hHE*̲| |*HμtrxD-x2#; 1oUIπ{'4^"}UjSھ"YruN^ň V`gz[4ےAxBׂW#v PpAһG.Ѽ& X+m kFV dKBҢ-41`WD(suzGhPWi-DWe%!~,2$)4z[tx9+$8Wt@)r0N?<=NT,_ VivwA-'tezwƎ Vu|He5 vh[pz 88\ ía|2)m~|tM=!!Рt <|2[-Q"@C͒3']"B%*;HwYNXb"+ ˇkv#h;dٓC8aUw/D~QrĖ3+wT=rMB Cf?Z}$^PSնCё zV{H^a>YΖxnrv rWhb'4OݖxX ښoczk5k2}z,G{E`*XO$](X1>@kSX?]sW8ɿ,KGژ2X舢x<9\@Oճ=$>҃,=@fHͮm<| KqYK ,0}$Wt9)s582XSk}bj2{V(&Gdߵ#UAʉç!V,N,z:&_Bc~'ptN j0nz< kNemܥRiS2keNɡ崆/ 7$K=ḟ&t{yO%;DP̜B}~u(`,Ԓ=ՒJVRIk#&O7)u@vw5 \E&y_+Q|6BՇoXXk4 ea@1H hWD(l)EeۂWag@r Joe,zۖ/<+;ϓq) Zl^:N|H}'Y+4 4mU%肪 ˋT&qiofyyq\5d!Bg!FGpwe@_ :~Х0@>n량# zڑ F(7r%qgzGdƮ傰Q_Z^d9bE XUk\)hf[W`TԪ,zp1nzQ4Y>Ky%-쒩B\TMt9asu&}!eG@@1yGj)Օ>,SM΅^ȅYWӴ7A9]S[ٟywsYx(+h=6Ou걞8Ժw