%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 13 0 R ] /Type /Pages /Count 10 >> 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 (2018) /EventType (Poster) /Description-Abstract (Real\055world image recognition is often challenged by the variability of visual styles including object textures\054 lighting conditions\054 filter effects\054 etc\056 Although these variations have been deemed to be implicitly handled by more training data and deeper networks\054 recent advances in image style transfer suggest that it is also possible to explicitly manipulate the style information\056 Extending this idea to general visual recognition problems\054 we present Batch\055Instance Normalization \050BIN\051 to explicitly normalize unnecessary styles from images\056 Considering certain style features play an essential role in discriminative tasks\054 BIN learns to selectively normalize only disturbing styles while preserving useful styles\056 The proposed normalization module is easily incorporated into existing network architectures such as Residual Networks\054 and surprisingly improves the recognition performance in various scenarios\056 Furthermore\054 experiments verify that BIN effectively adapts to completely different tasks like object classification and style transfer\054 by controlling the trade\055off between preserving and removing style variations\056 BIN can be implemented with only a few lines of code using popular deep learning frameworks\056) /Producer (PyPDF2) /Title (Batch\055Instance Normalization for Adaptively Style\055Invariant Neural Networks) /Date (2018) /ModDate (D\07220190218215039\05508\04700\047) /Published (2018) /Type (Conference Proceedings) /firstpage (2558) /Book (Advances in Neural Information Processing Systems 31) /Description (Paper accepted and presented at the Neural Information Processing Systems Conference \050http\072\057\057nips\056cc\057\051) /Editors (S\056 Bengio and H\056 Wallach and H\056 Larochelle and K\056 Grauman and N\056 Cesa\055Bianchi and R\056 Garnett) /Author (Hyeonseob Nam\054 Hyo\055Eun Kim) /lastpage (2567) >> endobj 3 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 4 0 obj << /Contents 14 0 R /Parent 1 0 R /Resources 15 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 38 0 R 39 0 R 40 0 R 41 0 R 42 0 R 43 0 R 44 0 R 45 0 R 46 0 R ] /Type /Page >> endobj 5 0 obj << /Contents 47 0 R /Parent 1 0 R /Resources 48 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 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 62 0 R 63 0 R 64 0 R 65 0 R 66 0 R 67 0 R ] /Type /Page >> endobj 6 0 obj << /Contents 68 0 R /Parent 1 0 R /Type /Page /Resources 69 0 R /MediaBox [ 0 0 612 792 ] >> endobj 7 0 obj << /Contents 106 0 R /Parent 1 0 R /Resources 107 0 R /Group 173 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 175 0 R 176 0 R 177 0 R 178 0 R ] /Type /Page >> endobj 8 0 obj << /Contents 179 0 R /Parent 1 0 R /Resources 180 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 222 0 R 223 0 R 224 0 R 225 0 R ] /Type /Page >> endobj 9 0 obj << /Contents 226 0 R /Parent 1 0 R /Resources 227 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 228 0 R 229 0 R 230 0 R 231 0 R 232 0 R 233 0 R 234 0 R 235 0 R 236 0 R 237 0 R 238 0 R 239 0 R 240 0 R 241 0 R ] /Type /Page >> endobj 10 0 obj << /Contents 242 0 R /Parent 1 0 R /Resources 243 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 428 0 R 429 0 R 430 0 R 431 0 R 432 0 R 433 0 R 434 0 R 435 0 R 436 0 R 437 0 R ] /Type /Page >> endobj 11 0 obj << /Contents 438 0 R /Parent 1 0 R /Resources 439 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 492 0 R 493 0 R 494 0 R 495 0 R 496 0 R 497 0 R 498 0 R 499 0 R 500 0 R ] /Type /Page >> endobj 12 0 obj << /Contents 501 0 R /Parent 1 0 R /Resources 502 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 503 0 R 504 0 R 505 0 R 506 0 R ] /Type /Page >> endobj 13 0 obj << /Contents 507 0 R /Parent 1 0 R /Type /Page /Resources 508 0 R /MediaBox [ 0 0 612 792 ] >> endobj 14 0 obj << /Length 3042 /Filter /FlateDecode >> stream xYr8+x"T4$4v/Vl*HnYŶ'V"D\^L}##2rQ$̢ui"Eu.zͻ7?Y$D,#$Syd")3=;;쏻lw;}l]Uⶣo