-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathreferences-full.bib
3125 lines (2650 loc) · 226 KB
/
references-full.bib
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
% This file was created with JabRef 2.10.
% Encoding: ISO8859_1
@Article{Adelson1992,
Title = {{Single lens stereo with a plenoptic camera}},
Author = {Adelson, E.H. and Wang, J.Y.a.},
Journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
Year = {1992},
Number = {2},
Pages = {99--106},
Volume = {14},
Doi = {10.1109/34.121783},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Adelson, Wang - 1992 - Single lens stereo with a plenoptic camera.pdf:pdf},
ISSN = {01628828},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=121783}
}
@InProceedings{ahmad1989scaling,
Title = {Scaling and generalization in neural networks: a case study},
Author = {Ahmad, Subutai and Tesauro, Gerald},
Booktitle = {Advances in neural information processing systems},
Year = {1989},
Pages = {160--168},
Owner = {yani},
Timestamp = {2017.09.13}
}
@Article{journals/neco/AmitG97,
Title = {{Shape Quantization And Recognition With Randomized Trees.}},
Author = {Amit, Yali and Geman, Donald},
Journal = {Neural Computation},
Year = {1997},
Number = {7},
Pages = {1545--1588},
Volume = {9},
Keywords = {dblp},
Url = {http://dblp.uni-trier.de/db/journals/neco/neco9.html{\#}AmitG97}
}
@InCollection{NIPS2014_5484,
Title = {{Do Deep Nets Really Need to be Deep?}},
Author = {Ba, Jimmy and Caruana, Rich},
Booktitle = {{Advances in Neural Information Processing Systems 27}},
Publisher = {Curran Associates, Inc.},
Year = {2014},
Editor = {Ghahramani, Z. and Welling, M. and Cortes, C. and Lawrence, N. D. and Weinberger, K. Q.},
Pages = {2654--2662},
Url = {http://papers.nips.cc/paper/5484-do-deep-nets-really-need-to-be-deep.pdf}
}
@InProceedings{Ba2013dothey,
Title = {{Do Deep Nets Really Need to be Deep ?}},
Author = {Ba, Lj and Caurana, R},
Booktitle = {{arXiv preprint arXiv:1312.6184}},
Year = {2013},
Pages = {1--6},
Volume = {2014},
Abstract = {Currently, deep neural networks are the state of the art on problems such as speech recognition and computer vision. In this extended abstract, we show that shal- low feed-forward networks can learn the complex functions previously learned by deep nets and achieve accuracies previously only achievable with deep models. Moreover, in some cases the shallow neural nets can learn these deep functions using a total number of parameters similar to the original deep model. We eval- uate our method on the TIMIT phoneme recognition task and are able to train shallow fully-connected nets that perform similarly to complex, well-engineered, deep convolutional architectures. Our success in training shallow neural nets to mimic deeper models suggests that there probably exist better algorithms for train- ing shallow feed-forward nets than those currently available.},
Archiveprefix = {arXiv},
Arxivid = {arXiv:1312.6184v5},
Doi = {10.1038/nature14539},
Eprint = {arXiv:1312.6184v5},
ISBN = {3135786504},
ISSN = {0028-0836},
Pmid = {26017442},
Url = {http://arxiv.org/abs/1312.6184}
}
@Article{Barron2012,
Title = {{Shape , Albedo , and Illumination from a Single Image of an Unknown Object}},
Author = {Barron, Jonathan T and Malik, Jitendra and Berkeley, U C},
Journal = {IEEE Conference on Computer Vision and Patern Recognition},
Year = {2012},
Pages = {334--341},
Abstract = {We address the problem of recovering shape, albedo, and illumination from a single grayscale image of an object, using shading as our primary cue. Because this problem is fundamentally underconstrained, we construct statistical models of albedo and shape, and define an optimization problem that searches for the most likely explanation of a single image. We present two priors on albedo which en- courage local smoothness and global sparsity, and three priors on shape which encourage flatness, outward-facing orientation at the occluding contour, and local smoothness. We present an optimization technique for using these pri- ors to recover shape, albedo, and a spherical harmonic model of illumination. Our model, which we call SAIFS (shape, albedo, and illumination from shading) produces reasonable results on arbitrary grayscale images taken in the real world, and outperforms all previous grayscale in- trinsic image-style algorithms on the MIT Intrinsic Images dataset.},
Doi = {10.1109/CVPR.2012.6247693},
ISBN = {9781467312288},
ISSN = {10636919},
Publisher = {IEEE},
Url = {http://ieeexplore.ieee.org/articleDetails.jsp?arnumber=6247693{&}contentType=Conference+Publications}
}
@Article{bartlett1997,
Title = {{For valid generalization, the size of the weights is more important than the size of the network}},
Author = {Bartlett, Peter L},
Journal = {Advances in neural information processing systems},
Year = {1997},
Pages = {134--140},
Publisher = {MORGAN KAUFMANN PUBLISHERS}
}
@InProceedings{Bastani2016,
Title = {{Measuring Neural Net Robustness with Constraints}},
Author = {Bastani, Osbert and Ioannou, Yani and Lampropoulos, Leonidas and Vytiniotis, Dimitrios and Nori, Aditya and Criminisi, Antonio},
Booktitle = {{Neural Information Processing Systems (NIPS), 2016}},
Year = {2016},
Address = {Barcelona, Spain},
Month = {dec},
Abstract = {Despite having high accuracy, neural nets have been shown to be susceptible to adversarial examples, where a small perturbation to an input can cause it to become mislabeled. We propose metrics for measuring the robustness of a neural net and devise a novel algorithm for approximating these metrics based on an encoding of robustness as a linear program. We show how our metrics can be used to evaluate the robustness of deep neural nets with experiments on the MNIST and CIFAR-10 datasets. Our algorithm generates more informative estimates of robustness metrics compared to estimates based on existing algorithms. Furthermore, we show how existing approaches to improving robustness "overfit" to adversarial examples generated using a specific algorithm. Finally, we show that our techniques can be used to additionally improve neural net robustness both according to the metrics that we propose, but also according to previously proposed metrics.},
Archiveprefix = {arXiv},
Arxivid = {1605.07262},
Eprint = {1605.07262},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Bastani et al. - 2016 - Measuring Neural Net Robustness with Constraints.pdf:pdf},
Keywords = {CNN; adversarial images; deep learning; robustness},
Mendeley-tags = {CNN,adversarial images,deep learning,robustness},
Url = {http://arxiv.org/abs/1605.07262}
}
@InProceedings{bastani2016measuring,
Title = {{Measuring Neural Net Robustness with Constraints}},
Author = {Bastani, Osbert and Ioannou, Yani and Lampropoulos, Leonidas and Vytiniotis, Dimitrios and Nori, Aditya and Criminisi, Antonio},
Booktitle = {{Neural Information Processing Systems (NIPS), 2016}},
Year = {2016}
}
@InProceedings{baum1989size,
Title = {{What size net gives valid generalization?}},
Author = {Baum, Eric B and Haussler, David},
Booktitle = {{Advances in neural information processing systems}},
Year = {1989},
Pages = {81--90}
}
@Misc{Beacco2003,
Title = {{A system for in situ measurements of road reflection properties}},
Author = {Beacco, D and Fiorentin, P and Rossi, G},
Year = {2003},
Abstract = {The characterization of the photometric properties of a road surface is of prime importance in the design of lighting plant and when the real vision condition should be determined by computer simulation. The measurement could be done in laboratory but the in situ measurement are very interested because it permit to test several zone on the road and there is no mechanical starch on the surface of the sample. This work describes an innovative portable system based on a CCD luminance meter able to obtain uncertainty comparable in traditional laboratory systems.},
Booktitle = {{Proceedings of the 20th IEEE Instrumentation Technology Conference Cat No03CH37412}},
Doi = {10.1109/IMTC.2003.1208001},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Beacco, Fiorentin, Rossi - 2003 - A system for in situ measurements of road reflection properties.pdf:pdf},
ISBN = {0780377052},
ISSN = {10915281},
Number = {May},
Pages = {1508--1512},
Publisher = {Ieee},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=1208001},
Volume = {2}
}
@Article{Bellia2002,
Title = {{Setting up a CCD photometer for lighting research and design}},
Author = {Bellia, L and Cesarano, A and Minichiello, F and Sibilio, S},
Journal = {Building and Environment},
Year = {2002},
Number = {11},
Pages = {1099--1106},
Volume = {37},
Abstract = {Recent availability of video-cameras with CCD-type sensors (charge coupled device) has proved to be particularly stimulating for all those applications requiring photometric measurements, above all for the measurement of luminance values related to the physical and technical qualities of a built environment. This method allows the instantaneous capture of an image, thus enabling collection of luminance values relating to the points of measurement; this in turn leads to the evaluation of luminance distribution and lighting levels of the surfaces that make up the environment. Setting up this system requires the following basic configuration: a photopic filter V($\lambda$), an optic interface, a computer equipped with an appropriate card for the capture and digitalisation of the acquired image (the grqqframe grabber) and, finally, suitable software for the processing of collected data. In this article a detailed description of this acquisition system is reported, and subsequently a report on the procedure adopted for its calibration so as to enable the capture of relevant photometric values. Final analysis and validation of results are carried out by means of field test. A case study of CCD photometerapplication has been then performed using a basic software tool autonomously developed to evaluate indoor lighting level; the luminance map of a diffuse light source has been used as grqqinput data for the developed software, and the grqqoutput data, i.e. illumination levels, have been then compared with measured values.},
Doi = {10.1016/S0360-1323(01)00093-2},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Bellia et al. - 2002 - Setting up a CCD photometer for lighting research and design.pdf:pdf},
ISSN = {03601323},
Keywords = {calculation; luminance; photometers; software code; video camera},
Publisher = {CIE Poland},
Url = {http://linkinghub.elsevier.com/retrieve/pii/S0360132301000932}
}
@InProceedings{Bengio2010labeltree,
Title = {{Label Embedding Trees for Large Multi-Class Tasks}},
Author = {Bengio, S and Weston, J and Grangier, D},
Booktitle = {{Conference and Workshop on Neural Information Processing Systems}},
Year = {2010}
}
@Article{bengio:ieeenn94,
Title = {{Learning Long-Term Dependencies With Gradient Descent Is Difficult}},
Author = {Bengio, Yoshua and Simard, Patrick and Frasconi, Paolo},
Journal = {IEEE Transactions on Neural Networks},
Year = {1994},
Number = {2},
Pages = {157--166},
Volume = {5},
Keywords = {nn}
}
@Book{Bishop1995,
Title = {{Neural Networks for Pattern Recognition}},
Author = {Bishop, Christopher M},
Publisher = {Oxford University Press},
Year = {1995},
Address = {Oxford},
Keywords = {imported}
}
@InCollection{Bottou2012sgdtricks,
Title = {{Stochastic Gradient Descent Tricks.}},
Author = {Bottou, L{\'e}on},
Booktitle = {{Neural Networks: Tricks of the Trade (2nd ed.)}},
Publisher = {Springer},
Year = {2012},
Editor = {Montavon, Gr{\'e}goire and Orr, Genevieve B and M{\"u}ller, Klaus-Robert},
Pages = {421--436},
Series = {{Lecture Notes in Computer Science}},
Volume = {7700},
ISBN = {978-3-642-35288-1},
Keywords = {dblp}
}
@Article{breiman2001random,
Title = {{Random Forests}},
Author = {Breiman, Leo},
Journal = {Machine Learning},
Year = {2001},
Pages = {5--32},
Volume = {45},
Keywords = {forests random}
}
@Article{breiman1996bagging,
Title = {{Bagging Predictors}},
Author = {Breiman, Leo},
Journal = {Machine Learning},
Year = {1996},
Number = {421},
Pages = {123--140},
Volume = {24},
Abstract = {Bagging predictors is a method for generating multiple versions of a pre-dictor and using these to get an aggregated predictor. The aggregation av-erages over the versions when predicting a numerical outcome and does a plurality v ote when predicting a class. The multiple versions are formed by making bootstrap replicates of the learning set and using these as new learning sets. Tests on real and simulated data sets using classiication and regression trees and subset selection in linear regression show that bagging can give substantial gains in accuracy. The vital element is the instability o f the prediction method. If perturbing the learning set can cause signiicant changes in the predictor constructed, then bagging can improve accuracy.},
Doi = {10.1007/BF00058655},
ISBN = {0885-6125},
ISSN = {0885-6125},
Keywords = {aggregation; averaging; bootstrap; combining},
Pmid = {17634459},
Publisher = {Springer}
}
@Book{breiman84,
Title = {{Classification and regression trees}},
Author = {Breiman, Leo and Friedman, Jerome H and Olshen, Richard A and Stone, Charles J},
Publisher = {CRC press},
Year = {1984},
Booktitle = {{CA: Wadsworth International Group}},
Doi = {10.1371/journal.pone.0015807},
ISBN = {978-0534980535},
ISSN = {19326203},
Pmid = {462029}
}
@Article{burges1998tutorial,
Title = {{A tutorial on support vector machines for pattern recognition}},
Author = {Burges, Christopher JC},
Journal = {Data mining and knowledge discovery},
Year = {1998},
Number = {2},
Pages = {121--167},
Volume = {2},
Publisher = {Springer}
}
@InProceedings{caruana2001overfitting,
Title = {{Overfitting in neural nets: Backpropagation, conjugate gradient, and early stopping}},
Author = {Caruana, Rich and Lawrence, Steve and Giles, C Lee},
Booktitle = {{Advances in neural information processing systems}},
Year = {2001},
Pages = {402--408}
}
@Article{castellano1997iterative,
Title = {{An iterative pruning algorithm for feedforward neural networks}},
Author = {Castellano, Giovanna and Fanelli, Anna Maria and Pelillo, Marcello},
Journal = {IEEE Transactions on Neural Networks},
Year = {1997},
Number = {3},
Pages = {519--531},
Volume = {8},
Owner = {yani},
Publisher = {IEEE},
Timestamp = {2017.09.13}
}
@InProceedings{Chang2012,
Title = {{Active Attentional Sampling for Speed-up of Background Subtraction}},
Author = {Chang, Hyung Jin and Jeong, Hawook and Choi, And Jin Young},
Booktitle = {{IEEE Conference on Computer Vision and Pattern Recognition}},
Year = {2012}
}
@InProceedings{Chen2015,
Title = {{Compressing Neural Networks with the Hashing Trick}},
Author = {Chen, Wenlin and Wilson, James T. and Tyree, Stephen and Weinberger, Kilian Q. and Chen, Yixin},
Booktitle = {{Proceedings of The 32nd International Conference on Machine Learning}},
Year = {2015},
Editor = {Bach, Francis R and Blei, David M},
Pages = {2285--2294},
Publisher = {JMLR.org},
Series = {{JMLR Proceedings}},
Volume = {37},
Abstract = {As deep nets are increasingly used in applications suited for mobile devices, a fundamental dilemma becomes apparent: the trend in deep learning is to grow models to absorb ever-increasing data set sizes; however mobile devices are designed with very little memory and cannot store such large models. We present a novel network architecture, HashedNets, that exploits inherent redundancy in neural networks to achieve drastic reductions in model sizes. HashedNets uses a low-cost hash function to randomly group connection weights into hash buckets, and all connections within the same hash bucket share a single parameter value. These parameters are tuned to adjust to the HashedNets weight sharing architecture with standard backprop during training. Our hashing procedure introduces no additional memory overhead, and we demonstrate on several benchmark data sets that HashedNets shrink the storage requirements of neural networks substantially while mostly preserving generalization performance.},
Archiveprefix = {arXiv},
Arxivid = {1504.04788},
Eprint = {1504.04788},
ISBN = {9781510810587},
Keywords = {dblp},
Url = {http://arxiv.org/abs/1504.04788}
}
@InProceedings{Ciresan2012,
Title = {{Multi-column deep neural networks for image classification}},
Author = {Ciresan, Dan and Meier, Ueli and Schmidhuber, J{\"u}rgen},
Booktitle = {{arXiv:1202.2745v1 [cs.CV]}},
Year = {2012},
Pages = {3642--3649},
Abstract = {Traditional methods of computer vision and machine learning cannot match human performance on tasks such as the recognition of handwritten digits or traffic signs. Our biologically plausible deep artificial neural network architectures can. Small (often minimal) receptive fields of convolutional winnertake-all neurons yield large network depth, resulting in roughly as many sparsely connected neural layers as found in mammals between retina and visual cortex. Only winner neurons are trained. Several deep neural columns become experts on inputs preprocessed in different ways; their predictions are averaged. Graphics cards allow for fast training. On the very competitive MNIST handwriting benchmark, our method is the first to achieve near-human performance. On a traffic sign recognition benchmark it outperforms humans by a factor of two. We also improve the state-of-the-art on a plethora of common image classification benchmarks.},
Archiveprefix = {arXiv},
Arxivid = {1202.2745},
Eprint = {1202.2745},
ISBN = {1467312266},
Url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.299.4060{&}rep=rep1{&}type=pdf{\%}5Cnhttp://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.299.4060{&}rank=5}
}
@InProceedings{decov,
Title = {{Reducing Overfitting in Deep Networks by Decorrelating Representations}},
Author = {Cogswell, Michael and Ahmed, Faruk and Girshick, Ross and Zitnick, Larry and Batra, Dhruv},
Booktitle = {{{International Conference on Learning Representations (ICLR 2016)}, San Jose, Puerto Rico}},
Year = {2016},
Month = may,
Archiveprefix = {arXiv},
Eprint = {1511.06068v4}
}
@InProceedings{Cogswell2016,
Title = {{Reducing Overfitting in Deep Networks by Decorrelating Representations.}},
Author = {Cogswell, Michael and Ahmed, Faruk and Girshick, Ross B and Zitnick, Larry and Batra, Dhruv},
Booktitle = {{International Conference on Learning Representations}},
Year = {2016}
}
@Article{criminisi2013decision,
Title = {{Decision Forests for Computer Vision and Medical Image Analysis}},
Author = {Criminisi, Antonio and Shotton, Jamie},
Year = {2013},
Publisher = {Springer Publishing Company, Incorporated}
}
@Article{Cryer1999,
Title = {{Shape-from-shading: a survey}},
Author = {Cryer, J.E. and Shah, M.},
Journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
Year = {1999},
Number = {8},
Pages = {690--706},
Volume = {21},
Doi = {10.1109/34.784284},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Cryer, Shah - 1999 - Shape-from-shading a survey.pdf:pdf},
ISSN = {01628828},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=784284}
}
@Article{Cucchiara2001,
Title = {{Improving shadow suppression in moving object detection with HSV color information}},
Author = {Cucchiara, R. and Crana, C. and Piccardi, M. and Prati, a. and Sirotti, S.},
Journal = {ITSC 2001. 2001 IEEE Intelligent Transportation Systems. Proceedings (Cat. No.01TH8585)},
Year = {2001},
Pages = {334--339},
Doi = {10.1109/ITSC.2001.948679},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Cucchiara et al. - 2001 - Improving shadow suppression in moving object detection with HSV color information.pdf:pdf},
ISBN = {0-7803-7194-1},
Publisher = {Ieee},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=948679}
}
@Article{Cucchiara2003,
Title = {{Detecting Moving Objects , Ghosts , and Shadows in Video Streams {\ae}}},
Author = {Cucchiara, Rita and Grana, Costantino and Piccardi, Massimo and Prati, Andrea},
Year = {2003},
Number = {10},
Pages = {1337--1342},
Volume = {25},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Cucchiara et al. - 2003 - Detecting Moving Objects , Ghosts , and Shadows in Video Streams {\ae}.pdf:pdf}
}
@InProceedings{lecun1989optimal,
Title = {{Optimal Brain Damage}},
Author = {LeCun, Yann and Denker, John S. and Solla, Sara A.},
Booktitle = {{Advances in Neural Information Processing Systems}},
Year = {1990},
Number = {1},
Pages = {598--605},
Volume = {2},
Abstract = {We have used information-theoretic ideas to derive a class of practical and nearly optimal schemes for adapting the size of a neural network. By removing unimportant weights from a network, several improvements can be expected: better generalization, fewer training examples required, and improved speed of learning and/or classification. The basic idea is to use second-derivative information to make a tradeoff between network complexity and training set error. Experiments confirm the usefulness of the methods on a real-world application.},
Archiveprefix = {arXiv},
Arxivid = {arXiv:1011.1669v3},
Eprint = {arXiv:1011.1669v3},
ISBN = {1558601007},
ISSN = {1098-6596},
Pmid = {25246403}
}
@Article{journals/mcss/Cybenko92,
Title = {{Approximation by superpositions of a sigmoid function}},
Author = {Cybenko, G},
Journal = {Mathematics of control, signals, and systems},
Year = {1989},
Number = {4},
Pages = {303--314},
Volume = {2},
Keywords = {dblp},
Url = {http://dblp.uni-trier.de/db/journals/mcss/mcss5.html{\#}Cybenko92}
}
@Book{damelin2011,
Title = {{The Mathematics of Signal Processing}},
Author = {Damelin, Steven B. and {Miller Jr}, Willard},
Publisher = {Cambridge University Press},
Year = {2012},
Address = {Cambridge},
Abstract = {Arising from courses taught by the authors, this largely self-contained treatment is ideal for mathematicians who are interested in applications or for students from applied fields who want to understand the mathematics behind their subject. Early chapters cover Fourier analysis, functional analysis, probability and linear algebra, all of which have been chosen to prepare the reader for the applications to come. The book includes rigorous proofs of core results in compressive sensing and wavelet convergence. Fundamental is the treatment of the linear system y=$\Phi$x in both finite and infinite dimensions. There are three possibilities: the system is determined, overdetermined or underdetermined, each with different aspects. The authors assume only basic familiarity with advanced calculus, linear algebra and matrix theory and modest familiarity with signal processing, so the book is accessible to students from the advanced undergraduate level. Many exercises are also included.},
Doi = {10.1017/CBO9781139003896},
ISBN = {9781107601048},
Pages = {462},
Pmid = {17238176},
Url = {http://www.amazon.com/Mathematics-Signal-Processing-Cambridge-Applied/dp/1107601045/ref=pd{\_}sim{_}sbs{_}b{_}4?ie=UTF8{&}refRID=0TKKM2SWXXJPAXKE5KWG}
}
@Article{Debevec2008,
Title = {{Recovering high dynamic range radiance maps from photographs}},
Author = {Debevec, PE and Malik, J},
Journal = {ACM SIGGRAPH 2008 classes},
Year = {2008},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Unknown - 1997 - Recovering High Dynamic Range Radiance Maps from Photographs Paul E . Debevec.pdf:pdf},
Url = {http://dl.acm.org/citation.cfm?id=1401174}
}
@Article{DeMenthon1990,
Title = {{New exact and approximate solutions of the three-point perspective problem}},
Author = {DeMenthon, D and Davis, LS},
Journal = {Robotics and Automation, 1990. {\ldots}},
Year = {1990},
Pages = {40--45},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/DeMenthon, Davis - 1990 - New exact and approximate solutions of the three-point perspective problem.pdf:pdf},
Url = {http://ieeexplore.ieee.org/xpls/abs{\_}all.jsp?arnumber=125943}
}
@Misc{DeMenthon1992,
Title = {{Exact and approximate solutions of the perspective-three-point problem}},
Author = {DeMenthon, D and Davis, L S},
Year = {1992},
Abstract = {Model-based pose estimation techniques that match image and model triangles require large numbers of matching operations in real-world applications. The authors show that by using approximations to perspective, 2D lookup tables can be built for each of the triangles of the models. An approximation called `weak perspective' has been applied previously to this problem; the authors consider two other perspective approximations: paraperspective and orthoperspective. These approximations produce lower errors for off-center image features than weak perspective},
Booktitle = {{IEEE Transactions on Pattern Analysis and Machine Intelligence}},
Doi = {10.1109/34.166625},
ISSN = {01628828},
Number = {11},
Pages = {1100--1105},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=166625},
Volume = {14}
}
@InProceedings{Deng2011fastbalanced,
Title = {{Fast and Balanced: Efficient Label Tree Learning for Large Scale Object Recognition}},
Author = {Deng, J and Satheesh, S and Berg, A C and Li, F.-F.},
Booktitle = {{Conference and Workshop on Neural Information Processing Systems}},
Year = {2011}
}
@InProceedings{Denil2013predicting,
Title = {{Predicting Parameters in Deep Learning}},
Author = {Denil, Misha and Shakibi, Babak and Dinh, Laurent and Ranzato, Marc'Aurelio and de Freitas, Nando},
Booktitle = {{Neural Information Processing Systems (NIPS)}},
Year = {2013},
Pages = {2148--2156},
Abstract = {We demonstrate that there is significant redundancy in the parameterization of several deep learning models. Given only a few weight values for each feature it is possible to accurately predict the remaining values. Moreover, we show that not only can the parameter values be predicted, but many of them need not be learned at all. We train several different architectures by learning only a small number of weights and predicting the rest. In the best case we are able to predict more than 95{\%} of the weights of a network without any drop in accuracy.},
Archiveprefix = {arXiv},
Arxivid = {1306.0543},
Eprint = {1306.0543},
Url = {http://papers.nips.cc/paper/5025-predicting-parameters-in-deep-learning}
}
@Article{denker1987large,
Title = {{Large automatic learning, rule extraction, and generalization}},
Author = {Denker, John and Schwartz, Daniel and Wittner, Ben and Solla, Sara and Howard, Richard and Jackel, Lawrence and Hopfield, John},
Journal = {Complex systems},
Year = {1987},
Number = {5},
Pages = {877--922},
Volume = {1}
}
@InProceedings{Denton2014efficient,
Title = {{Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation}},
Author = {Denton, Emily and Zaremba, Wojciech and Bruna, Joan and LeCun, Yann and Fergus, Rob},
Booktitle = {{arXiv}},
Year = {2014},
Number = {1},
Pages = {1--11},
Abstract = {We present techniques for speeding up the test-time evaluation of large convolutional networks, designed for object recognition tasks. These models deliver impressive accuracy but each image evaluation requires millions of floating point operations, making their deployment on smartphones and Internet-scale clusters problematic. The computation is dominated by the convolution operations in the lower layers of the model. We exploit the linear structure present within the convolutional filters to derive approximations that significantly reduce the required computation. Using large state-of-the-art models, we demonstrate we demonstrate speedups of convolutional layers on both CPU and GPU by a factor of 2x, while keeping the accuracy within 1{\%} of the original model.},
Archiveprefix = {arXiv},
Arxivid = {1404.0736},
Eprint = {1404.0736},
ISSN = {10495258},
Url = {http://arxiv.org/abs/1404.0736}
}
@Article{Drew,
Title = {{Photometric stereo without multiple images 1 INTRODUCTION}},
Author = {Drew, Mark S},
Number = {604},
Pages = {369--380},
Volume = {3016},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Drew - Unknown - Photometric stereo without multiple images 1 INTRODUCTION.pdf:pdf},
Keywords = {based vision; color; dichromatic model; lambertian; neutral interface; physics; reflectance; shape; shape representation}
}
@Article{Edelman1998,
Title = {{The geometry of algorithms with orthogonality constraints}},
Author = {Edelman, A and Arias, TA},
Journal = {Arxiv preprint physics/9806030},
Year = {1998},
Archiveprefix = {arXiv},
Arxivid = {arXiv:physics/9806030v1},
Eprint = {9806030v1},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Edelman, Arias - 1998 - The geometry of algorithms with orthogonality constraints.pdf:pdf},
Keywords = {15a18; 49m07; 49m15; 51f20; 53b20; 65f15; 81v55; ams subject classifications; conjugate gradient; eigenvalue optimization; eigenvalues and eigenvectors; electronic structures computation; grassmann manifold; invariant subspace; newton; orthogonality constraints; programming; rayleigh quotient iteration; reduced gradient method; s method; sequential quadratic; stiefel manifold; subspace tracking},
Primaryclass = {arXiv:physics},
Url = {http://arxiv.org/abs/physics/9806030}
}
@InProceedings{Fahlman1989,
Title = {{The Cascade-Correlation Learning Architecture}},
Author = {Fahlmann, S E and Lebiere, C},
Booktitle = {{Advances in Neural Information Processing Systems 2}},
Year = {1990},
Editor = {Touretzky, David S},
Pages = {524--532},
Publisher = {Morgan Kaufmann},
Doi = {10.1190/1.1821929},
ISBN = {1558601007},
ISSN = {10459227},
Pmid = {220943591}
}
@Misc{Fleck1995,
Title = {{Perspective projection: the wrong imaging model}},
Author = {Fleck, Margaret M},
Year = {1995},
Abstract = {Perspective projection is generally accepted as the ideal model of image formation. Many recent algorithms, and many recent judgements about the relative merits of different algorithms, depend on this assumption. However, perspective projection represents only the front half of the viewing sphere and it distorts the shape and intensity of objects unless they lie near the optical axis. It is only one of several projections used in lens design and it does not accurately model the behavior of many real lenses. It works well only for narrow-angle images. This paper surveys the properties of several alternative models of image formation. A model based on stereographic projection of the viewing sphere is shown to be a better general-purpose imaging model than perspective projection. The new model can represent wider fields of view and more closely approximates real wide-angle lenses. It preserves a suitable range of shape properties, including local symmetries. It approximates narrow-angl...},
Booktitle = {{Research report}},
Pages = {95--01},
Publisher = {University of Iowa},
Url = {http://www.cs.illinois.edu/{~}mfleck/my-papers/stereographic-TR.pdf}
}
@InCollection{Hertzmann2005,
Title = {{Radiometry and Reflection}},
Author = {Fleet, David and Hertzmann, Aaron},
Year = {2005},
Pages = {76--91},
File = {:home/yani/Documents/LN12{\_}Lighting.pdf:pdf}
}
@Misc{fodor2002survey,
Title = {{A survey of dimension reduction techniques}},
Author = {Fodor, I K},
Year = {2002},
Abstract = {This paper, we assume that we have n observations, each being a realization of the p- dimensional random variable x = (x 1 , . . . , x p with mean E(x) = = 1 , . . . , p and covariance matrix E(x )(x = pp . We denote such an observation matrix by X = i,j : 1 p, 1 n. If i and i = (i,i) denote the mean and the standard deviation of the ith random variable, respectively, then we will often standardize the observations x i,j by (x i,j i i , where i = x i = 1/n j=1 x i,j , and i = 1/n j=1 (x i,j x i},
Booktitle = {{Center for Applied Scientific Computing Lawrence Livermore National Laboratory}},
Doi = {10.2172/15002155},
Pages = {1--18},
Publisher = {Technical Report UCRL-ID-148494, Lawrence Livermore National Laboratory},
Url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.8.5098},
Volume = {9}
}
@Article{frean1990upstart,
Title = {The upstart algorithm: A method for constructing and training feedforward neural networks},
Author = {Frean, Marcus},
Journal = {Neural computation},
Year = {1990},
Number = {2},
Pages = {198--209},
Volume = {2},
Owner = {yani},
Publisher = {MIT Press},
Timestamp = {2017.09.13}
}
@Article{fukushima2013artificial,
Title = {{Artificial vision by multi-layered neural networks: Neocognitron and its advances}},
Author = {Fukushima, Kunihiko},
Journal = {Neural Networks},
Year = {2013},
Pages = {103--119},
Volume = {37},
Abstract = {The neocognitron is a neural network model proposed by. Fukushima (1980). Its architecture was suggested by neurophysiological findings on the visual systems of mammals. It is a hierarchical multi-layered network. It acquires the ability to robustly recognize visual patterns through learning. Although the neocognitron has a long history, modifications of the network to improve its performance are still going on. For example, a recent neocognitron uses a new learning rule, named add-if-silent, which makes the learning process much simpler and more stable. Nevertheless, a high recognition rate can be kept with a smaller scale of the network. Referring to the history of the neocognitron, this paper discusses recent advances in the neocognitron. We also show that various new functions can be realized by, for example, introducing top-down connections to the neocognitron: mechanism of selective attention, recognition and completion of partly occluded patterns, restoring occluded contours, and so on. {\textcopyright} 2012 Elsevier Ltd.},
Doi = {10.1016/j.neunet.2012.09.016},
ISSN = {08936080},
Keywords = {Artificial vision; Bottom-up and top-down; Hierarchical network; Modeling neural networks; Neocognitron},
Pmid = {23098752},
Publisher = {Elsevier}
}
@Article{Fuk80,
Title = {{Neocognitron: A self-organizing neural network model for a mechanish of pattern recognition unaffected by shifts in position}},
Author = {Fukushima, K},
Journal = {Biological Cybernetics},
Year = {1980},
Pages = {193--202},
Volume = {36},
Keywords = {deep fukushima learning neocognitron networks neur}
}
@InProceedings{Gal2016Dropout,
Title = {{Dropout as a {B}ayesian Approximation: Representing Model Uncertainty in Deep Learning}},
Author = {Gal, Yarin and Ghahramani, Zoubin},
Booktitle = {{Proceedings of the 33rd International Conference on Machine Learning (ICML-16)}},
Year = {2016}
}
@Article{Geiger2012,
Title = {{Are we ready for autonomous driving? the kitti vision benchmark suite}},
Author = {Geiger, Andreas and Lenz, Philip and Urtasun, Raquel},
Journal = {Computer Vision and},
Year = {2012},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Geiger, Lenz, Urtasun - 2012 - Are we ready for autonomous driving the kitti vision benchmark suite.pdf:pdf},
Url = {http://h1997453.stratoserver.net/publications/cvpr12.pdf}
}
@Article{giles1987learning,
Title = {{Learning, invariance, and generalization in high-order neural networks}},
Author = {Giles, C Lee and Maxwell, Tom},
Journal = {Applied optics},
Year = {1987},
Number = {23},
Pages = {4972--4978},
Volume = {26},
Publisher = {Optical Society of America}
}
@InProceedings{girshick2015deformable,
Title = {{Deformable Part Models are Convolutional Neural Networks}},
Author = {Girshick, Ross and Iandola, Forrest and Darrell, Trevor and Malik, Jitendra},
Booktitle = {{Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition}},
Year = {2015},
Pages = {437--446}
}
@InProceedings{glorot2010understanding,
Title = {{Understanding the difficulty of training deep feedforward neural networks}},
Author = {Glorot, Xavier and Bengio, Yoshua},
Booktitle = {{Proceedings of the 13th International Conference on Artificial Intelligence and Statistics (AISTATS)}},
Year = {2010},
Pages = {249--256},
Volume = {9},
Abstract = {Whereas before 2006 it appears that deep multilayer neural networks were not successfully trained, since then several algorithms have been shown to successfully train them, with experimental results showing the superiority of deeper vs less deep architectures. All these experimental results were obtained with new initialization or training mechanisms. Our objective here is to understand better why standard gradient descent from random initialization is doing so poorly with deep neural networks, to better understand these recent relative successes and help design better algorithms in the future. We first observe the influence of the non-linear activations functions. We find that the logistic sigmoid activation is unsuited for deep networks with random initialization because of its mean value, which can drive especially the top hidden layer into saturation. Surprisingly, we find that saturated units can move out of saturation by themselves, albeit slowly, and explaining the plateaus sometimes seen when training neural networks. We find that a new non-linearity that saturates less can often be beneficial. Finally, we study how activations and gradients vary across layers and during training, with the idea that training may be more difficult when the singular values of the Jacobian associated with each layer are far from 1. Based on these considerations, we propose a new initialization scheme that brings substantially faster convergence.},
ISSN = {15324435},
Url = {http://machinelearning.wustl.edu/mlpapers/paper{\_}files/AISTATS2010{_}GlorotB10.pdf}
}
@Article{Golovinskiy2009,
Title = {{Min-cut based segmentation of point clouds}},
Author = {Golovinskiy, Aleksey and Funkhouser, Thomas},
Journal = {2009 IEEE 12th International Conference on Computer Vision Workshops ICCV Workshops},
Year = {2009},
Pages = {39--46},
Volume = {150},
Abstract = {We present a min-cut based method of segmenting objects in point clouds. Given an object location, our method builds a k-nearest neighbors graph, assumes a background prior, adds hard foreground (and optionally background) constraints, and finds the min-cut to compute a foreground-background segmentation. Our method can be run fully automatically, or interactively with a user interface. We test our system on an outdoor urban scan, quantitatively evaluate our algorithm on a test set of about 1000 objects, and compare to several alternative approaches.},
Doi = {10.1109/ICCVW.2009.5457721},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Golovinskiy, Funkhouser - 2009 - Min-cut based segmentation of point clouds.pdf:pdf},
ISBN = {9781424444427},
Publisher = {Ieee},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=5457721}
}
@Book{goodfellow2016deep,
Title = {{Deep learning}},
Author = {Goodfellow, Ian and Bengio, Yoshua and Courville, Aaron},
Publisher = {MIT press},
Year = {2016}
}
@InProceedings{goodfellow2013maxout,
Title = {{Maxout Networks}},
Author = {Goodfellow, Ian J and Warde-Farley, David and Mirza, Mehdi and Courville, Aaron and Bengio, Yoshua},
Booktitle = {{Proceedings of the 30th International Conference on Machine Learning (ICML)}},
Year = {2013},
Pages = {1319--1327},
Volume = {28},
Abstract = {We consider the problem of designing mod-els to leverage a recently introduced ap-proximate model averaging technique called dropout. We define a simple new model called maxout (so named because its output is the max of a set of inputs, and because it is a nat-ural companion to dropout) designed to both facilitate optimization by dropout and im-prove the accuracy of dropout's fast approxi-mate model averaging technique. We empir-ically verify that the model successfully ac-complishes both of these tasks. We use max-out and dropout to demonstrate state of the art classification performance on four bench-mark datasets: MNIST, CIFAR-10, CIFAR-100, and SVHN.},
Archiveprefix = {arXiv},
Arxivid = {1302.4389},
Eprint = {1302.4389}
}
@Article{gorodkin1993quantitative,
Title = {A quantitative study of pruning by optimal brain damage},
Author = {Gorodkin, Jan and Hansen, Lars Kai and Krogh, Anders and Svarer, Claus and Winther, Ole},
Journal = {International journal of neural systems},
Year = {1993},
Number = {02},
Pages = {159--169},
Volume = {4},
Owner = {yani},
Publisher = {World Scientific},
Timestamp = {2017.09.13}
}
@Article{Gortler1996,
Title = {{The lumigraph}},
Author = {Gortler, Steven J. and Grzeszczuk, Radek and Szeliski, Richard and Cohen, Michael F.},
Journal = {Proceedings of the 23rd annual conference on Computer graphics and interactive techniques - SIGGRAPH '96},
Year = {1996},
Pages = {43--54},
Address = {New York, New York, USA},
Doi = {10.1145/237170.237200},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Gortler et al. - 1996 - The lumigraph.pdf:pdf},
ISBN = {0897917464},
Publisher = {ACM Press},
Url = {http://portal.acm.org/citation.cfm?doid=237170.237200}
}
@InProceedings{Gupta2015,
Title = {{Deep Learning with Limited Numerical Precision}},
Author = {Gupta, Suyog and Agrawal, Ankur and Gopalakrishnan, Kailash and Narayanan, Pritish},
Booktitle = {{Proceedings of the 32nd International Conference on Machine Learning (ICML-15)}},
Year = {2015},
Editor = {Blei, David and Bach, Francis},
Pages = {1737--1746},
Publisher = {JMLR Workshop and Conference Proceedings}
}
@Misc{1502.02551v1,
Title = {{Deep Learning with Limited Numerical Precision}},
Author = {Gupta, Suyog and Agrawal, Ankur and Gopalakrishnan, Kailash and Narayanan, Pritish},
Month = feb,
Year = {2015},
Abstract = {Training of large-scale deep neural networks is often constrained by the available computational resources. We study the effect of limited preci- sion data representation and computation on neu- ral network training. Within the context of low- precision fixed-point computations, we observe the rounding scheme to play a crucial role in de- termining the network's behavior during train- ing. Our results show that deep networks can be trained using only 16-bit wide fixed-point num- ber representation when using stochastic round- ing, and incur little to no degradation in the classification accuracy. We also demonstrate an energy-efficient hardware accelerator that imple- ments low-precision fixed-point arithmetic with stochastic rounding.},
Annote = {published = 2015-02-09T16:37:29Z, updated = 2015-02-09T16:37:29Z, 10 pages, 6 figures, 1 table},
Archiveprefix = {arXiv},
Arxivid = {1502.02551},
Booktitle = {{Proceedings of the 32nd International Conference on Machine Learning (ICML-15)}},
Doi = {10.1109/72.80206},
Eprint = {1502.02551},
ISBN = {9781510810587},
ISSN = {19410093},
Pages = {1737--1746},
Pmid = {18282824},
Url = {http://jmlr.org/proceedings/papers/v37/gupta15.pdf}
}
@InProceedings{hypernetworks,
Title = {{HyperNetworks}},
Author = {Ha, David and Dai, Andrew and Le, Quoc V.},
Booktitle = {{International Conference on Learning Representations (ICLR), Toulon, France}},
Year = {2017}
}
@Article{han2015deep,
Title = {{Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding}},
Author = {Han, Song and Mao, Huizi and Dally, William J},
Year = {2016},
Booktitle = {{International Conference on Learning Representations (ICLR), San Juan, Puerto Rico}}
}
@InProceedings{han2016dsd,
Title = {{Dsd: Dense-sparse-dense training for deep neural networks}},
Author = {Han, Song and Pool, Jeff and Narang, Sharan and Mao, Huizi and Gong, Enhao and Tang, Shijian and Elsen, Erich and Vajda, Peter and Paluri, Manohar and Tran, John and others},
Booktitle = {{International Conference on Learning Representations (ICLR), Toulon, France}},
Year = {2017}
}
@InProceedings{han2015learning,
Title = {{Learning both weights and connections for efficient neural network}},
Author = {Han, Song and Pool, Jeff and Tran, John and Dally, William},
Booktitle = {Advances in Neural Information Processing Systems},
Year = {2015},
Pages = {1135--1143},
Owner = {yani},
Timestamp = {2017.09.13}
}
@Article{journals/iandc/HancockJLT96,
Title = {{Lower Bounds on Learning Decision Lists and Trees}},
Author = {Hancock, Thomas and Jiang, Tao and Li, Ming and Tromp, John},
Journal = {Information and Computation},
Year = {1996},
Number = {2},
Pages = {114--122},
Volume = {126},
Abstract = {k-Decision lists and decision trees play important roles in learning theory as well as in practical learning systems.k-Decision lists generalize classes such as monomials,k-DNF, andk-CNF, and like these subclasses they are polynomially PAC-learnable [R. Rivest,Mach. Learning2(1987), 229--246]. This leaves open the question of whetherk-decision lists can be learned as efficiently ask-DNF. We answer this question negatively in a certain sense, thus disproving a claim in a popular textbook [M. Anthony and N. Biggs, ``Computational Learning Theory,'' Cambridge Univ. Press, Cambridge, UK, 1992]. Decision trees, on the other hand, are not even known to be polynomially PAC-learnable, despite their widespread practical application. We will show that decision trees are not likely to be efficiently PAC-learnable. We summarize our specific results. The following problems cannot be approximated in polynomial time within a factor of 2log$\delta$ nfor any$\delta${\textless}1, unlessNP⊂DTIME[2polylog n]: a generalized set cover,k-decision lists,k-decision lists by monotone decision lists, and decision trees. Decision lists cannot be approximated in polynomial time within a factor ofn$\delta$, for some constant$\delta${\textgreater}0, unlessNP=P. Also,k-decision lists withl0--1 alternations cannot be approximated within a factor logl nunlessNP⊂DTIME[nO(log log n)] (providing an interesting comparison to the upper bound obtained by A. Dhagat and L. Hellerstein [in``FOCS '94,'' pp. 64--74]).},
Doi = {10.1006/inco.1996.0040},
ISBN = {3540590420},
ISSN = {0890-5401},
Url = {http://www.sciencedirect.com/science/article/pii/S0890540196900401{\%}5Cnhttp://www.sciencedirect.com/science/article/pii/S0890540196900401/pdf?md5=59bdd8c077309262836d57b76a5a5577{&}pid=1-s2.0-S0890540196900401-main.pdf}
}
@Misc{Hanmandlu2000,
Title = {{Depth estimation from a sequence of images using spherical projection}},
Author = {Hanmandlu, M and Shantaram, V and Sudheer, K},
Year = {2000},
Abstract = {A recursive estimation of depth from a sequence of images is proposed. Using the spherical projection, a simple equation is derived that relates image motion with the object motion. This equation is reformulated into a dynamical state space model for which Kalman filter can be easily applied to yield the estimate of depth. Point correspondences have been used to obtain feature points and the motion parameters are assumed to be known. The results are illustrated on a real object},
Booktitle = {{Proceedings of International Conference on Robotics and Automation}},
Doi = {10.1109/ITCC.2000.844211},
ISBN = {0769505406},
Number = {April},
Pages = {2264--2269},
Publisher = {Ieee},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=619298},
Volume = {3}
}
@InProceedings{hanson1989comparing,
Title = {{Comparing biases for minimal network construction with back-propagation}},
Author = {Hanson, Stephen Jos{\'e} and Pratt, Lorien Y},
Booktitle = {{Advances in neural information processing systems}},
Year = {1989},
Pages = {177--185}
}
@Article{Happel1994,
Title = {{Design and evolution of modular neural network architectures.}},
Author = {Happel, Bart L M and Murre, Jacob M J},
Journal = {Neural Networks},
Year = {1994},
Number = {6-7},
Pages = {985--1004},
Volume = {7}
}
@Article{Haralick1989,
Title = {{Monocular vision using inverse perspective projection geometry: analytic relations}},
Author = {Haralick, R M},
Journal = {Proceedings CVPR 89 IEEE Computer Society Conference on Computer Vision and Pattern Recognition},
Year = {1989},
Pages = {370--378},
Volume = {10},
Doi = {10.1109/CVPR.1989.37874},
ISBN = {081861918X},
Publisher = {IEEE Comput. Soc. Press},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=37874}
}
@InProceedings{Hardt2015,
Title = {{Train faster, generalize better: Stability of stochastic gradient descent}},
Author = {Hardt, Moritz and Recht, Benjamin and Singer, Yoram},
Booktitle = {{Proceedings of the 33rd International Conference on Machine Learning (ICML 2016)}},
Year = {2015},
Address = {New York, New York, USA},
Pages = {1--24},
Abstract = {We show that any model trained by a stochastic gradient method with few iterations has vanishing generalization error. We prove this by showing the method is algorithmically stable in the sense of Bousquet and Elisseeff. Our analysis only employs elementary tools from convex and continuous optimization. Our results apply to both convex and non-convex optimization under standard Lipschitz and smoothness assumptions. Applying our results to the convex case, we provide new explanations for why multiple epochs of stochastic gradient descent generalize well in practice. In the nonconvex case, we provide a new interpretation of common practices in neural networks, and provide a formal rationale for stability-promoting mechanisms in training large, deep models. Conceptually, our findings underscore the importance of reducing training time beyond its obvious benefit.},
Archiveprefix = {arXiv},
Arxivid = {1509.01240},
Eprint = {1509.01240},
ISBN = {9781510829008}
}
@Article{Hasinoff2010,
Title = {{Noise-optimal capture for high dynamic range photography}},
Author = {Hasinoff, Samuel W. and Durand, Fredo and Freeman, William T.},
Journal = {2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition},
Year = {2010},
Month = {jun},
Pages = {553--560},
Doi = {10.1109/CVPR.2010.5540167},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Hasinoff, Durand, Freeman - 2010 - Noise-optimal capture for high dynamic range photography.pdf:pdf},
ISBN = {978-1-4244-6984-0},
Publisher = {Ieee},
Url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=5540167}
}
@Book{haykin1994neural,
Title = {Neural networks: a comprehensive foundation},
Author = {Haykin, Simon},
Publisher = {Prentice Hall PTR},
Year = {1994}
}
@InProceedings{He2012,
Title = {{Incremental Gradient on the Grassmannian for Online Foreground and Background Separation in Subsampled Video}},
Author = {He, Jun and Balzano, Laura and Szlam, Arthur},
Year = {2012},
Pages = {1568--1575},
File = {:home/yani/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/He, Balzano, Szlam - 2012 - Incremental Gradient on the Grassmannian for Online Foreground and Background Separation in Subsampled Video.pdf:pdf},
ISBN = {9781467312288}
}
@InProceedings{he2015convolutional,
Title = {{Convolutional Neural Networks at Constrained Time Cost}},
Author = {He, Kaiming and Sun, Jian},
Booktitle = {{Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition}},
Year = {2015},
Pages = {5353--5360}
}
@Article{He2016,
Title = {{Identity Mappings in Deep Residual Networks}},
Author = {He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
Journal = {arXiv preprint},
Year = {2016},
Pages = {1--15},
Volume = {abs/1603.0},
Abstract = {Deep residual networks have emerged as a family of extremely deep architectures showing compelling accuracy and nice convergence behaviors. In this paper, we analyze the propagation formulations behind the residual building blocks, which suggest that the forward and backward signals can be directly propagated from one block to any other block, when using identity mappings as the skip connections and after-addition activation. A series of ablation experiments support the importance of these identity mappings. This motivates us to propose a new residual unit, which further makes training easy and improves generalization. We report improved results using a 1001-layer ResNet on CIFAR-10/100, and a 200-layer ResNet on ImageNet.},
Archiveprefix = {arXiv},
Arxivid = {1603.05027},
Eprint = {1603.05027},
Url = {http://arxiv.org/abs/1603.05027}
}
@Article{He2015,
Title = {{Deep Residual Learning for Image Recognition}},
Author = {He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
Journal = {Arxiv.Org},
Year = {2015},
Number = {3},
Pages = {171--180},
Volume = {7},
Abstract = {Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learn- ing residual functions with reference to the layer inputs, in- stead of learning unreferenced functions. We provide com- prehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8× deeper than VGG nets [41] but still having lower complex- ity. An ensemble of these residual nets achieves 3.57{\%} error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our ex- tremely deep representations, we obtain a 28{\%} relative im- provement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC {\&} COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet local- ization, COCO detection, and COCO segmentation.},
Archiveprefix = {arXiv},
Arxivid = {1512.03385},
Doi = {10.3389/fpsyg.2013.00124},
Eprint = {1512.03385},
ISBN = {978-1-4673-6964-0},
ISSN = {1664-1078},
Keywords = {deep learning; denoising auto-encoder; image denoising},
Pmid = {23554596},
Url = {http://arxiv.org/pdf/1512.03385v1.pdf}
}
@InProceedings{He2015b,
Title = {{Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification}},
Author = {He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
Booktitle = {{IEEE Conference on Computer Vision and Patern Recognition (ICCV)}},
Year = {2015},
Pages = {1026--1034},
Publisher = {IEEE},
Abstract = {Rectified activation units (rectifiers) are essential for state-of-the-art neural networks. In this work, we study rectifier neural networks for image classification from two aspects. First, we propose a Parametric Rectified Linear Unit (PReLU) that generalizes the traditional rectified unit. PReLU improves model fitting with nearly zero extra com-putational cost and little overfitting risk. Second, we derive a robust initialization method that particularly considers the rectifier nonlinearities. This method enables us to train extremely deep rectified models directly from scratch and to investigate deeper or wider network architectures. Based on the learnable activation and advanced initialization, we achieve 4.94{\%} top-5 test error on the ImageNet 2012 clas-sification dataset. This is a 26{\%} relative improvement over the ILSVRC 2014 winner (GoogLeNet, 6.66{\%} [33]). To our knowledge, our result is the first 1 to surpass the reported human-level performance (5.1{\%}, [26]) on this dataset.},
Archiveprefix = {arXiv},
Arxivid = {1502.01852},
Doi = {10.1109/ICCV.2015.123},
Eprint = {1502.01852},