%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) /Language (en\055US) /Created (2009) /Description-Abstract (We address the problem of learning classifiers when observations have multiple views\054 some of which may not be observed for all examples\056 We assume the existence of view generating functions which may complete the missing views in an approximate way\056 This situation corresponds for example to learning text classifiers from multilingual collections where documents are not available in all languages\056 In that case\054 Machine Translation \050MT\051 systems may be used to translate each document in the missing languages\056 We derive a generalization error bound for classifiers learned on examples with multiple artificially created views\056 Our result uncovers a trade\055off between the size of the training set\054 the number of views\054 and the quality of the view generating functions\056 As a consequence\054 we identify situations where it is more interesting to use multiple views for learning instead of classical single view learning\056 An extension of this framework is a natural way to leverage unlabeled multi\055view data in semi\055supervised learning\056 Experimental results on a subset of the Reuters RCV1\057RCV2 collections support our findings by showing that additional views obtained from MT may significantly improve the classification performance in the cases identified by our trade\055off\056) /Producer (Python PDF Library \055 http\072\057\057pybrary\056net\057pyPdf\057) /Title (Learning from Multiple Partially Observed Views \055 an Application to Multilingual Text Categorization) /Date (2009) /Type (Conference Proceedings) /firstpage (28) /Book (Advances in Neural Information Processing Systems 22) /Description (Paper accepted and presented at the Neural Information Processing Systems Conference \050http\072\057\057nips\056cc\057\051) /Editors (Y\056 Bengio and D\056 Schuurmans and J\056D\056 Lafferty and C\056K\056I\056 Williams and A\056 Culotta) /Author (Massih Amini\054 Nicolas Usunier\054 Cyril Goutte) /lastpage (36) >> endobj 3 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 4 0 obj << /Parent 1 0 R /Rotate 0 /Contents 13 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 14 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 5 0 obj << /Parent 1 0 R /Rotate 0 /Contents 22 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 23 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 6 0 obj << /Parent 1 0 R /Rotate 0 /Contents 69 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 70 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 7 0 obj << /Parent 1 0 R /Rotate 0 /Contents 77 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 78 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 8 0 obj << /Parent 1 0 R /Rotate 0 /Contents 79 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 80 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 9 0 obj << /Parent 1 0 R /Rotate 0 /Contents 81 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 82 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 10 0 obj << /Parent 1 0 R /Rotate 0 /Contents 83 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 84 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 11 0 obj << /Parent 1 0 R /Rotate 0 /Contents 91 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 92 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 12 0 obj << /Parent 1 0 R /Rotate 0 /Contents 94 0 R /Resources << /ProcSet [ /PDF /Text ] /Font 95 0 R >> /MediaBox [ 0 0 612 792 ] /Type /Page >> endobj 13 0 obj << /Length 6083 /Filter /FlateDecode >> stream x=r%sWߛ+a_n(