-
Notifications
You must be signed in to change notification settings - Fork 183
/
Copy pathindex.html
1083 lines (937 loc) · 61 KB
/
index.html
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
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<title>Yunhe Wang's Homepage</title>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<meta name="description" content="Yunhe Wang is currently a senior researcher at Huawei Noah's Ark Lab">
<meta name="keywords" content="Yunhe Wang, 王云鹤, wangyunhe, Yunhe, Wang, Deep Learning, Huawei, PKU, Computer, Vision">
<meta name="author" content="Yunhe Wang" />
<link rel="stylesheet" href="w3.css">
<style>
.w3-sidebar a {font-family: "Roboto", sans-serif}
body,h1,h2,h3,h4,h5,h6,.w3-wide {font-family: "Montserrat", sans-serif;}
</style>
<link rel="icon" type="image/png" href="images/icons.png">
<!--
<script src="jquery.min.js"></script>
<script>
$(document).ready(function(){
// Add smooth scrolling to all links
$("a").on('click', function(event) {
// Make sure this.hash has a value before overriding default behavior
if (this.hash !== "") {
// Prevent default anchor click behavior
event.preventDefault();
// Store hash
var hash = this.hash;
// Using jQuery's animate() method to add smooth page scroll
// The optional number (800) specifies the number of milliseconds it takes to scroll to the specified area
$('html, body').animate({
scrollTop: $(hash).offset().top
}, 800, function(){
// Add hash (#) to URL when done scrolling (default click behavior)
window.location.hash = hash;
});
} // End if
});
});
</script>
//-->
</head>
<body class="w3-content" style="max-width:1000px">
<!-- Sidebar/menu -->
<nav class="w3-sidebar w3-bar-block w3-black w3-collapse w3-top w3-right" style="z-index:3;width:150px" id="mySidebar">
<div class="w3-container w3-display-container w3-padding-16">
<h3><b>YUNHE</b></h3>
</div>
<div class="w3-padding-64 w3-text-light-grey w3-large" style="font-weight:bold">
<a href="#home" class="w3-bar-item w3-button">Home</a>
<a href="#news" class="w3-bar-item w3-button">News</a>
<a href="#projects" class="w3-bar-item w3-button">Projects</a>
<a href="#talks" class="w3-bar-item w3-button">Talks</a>
<a href="#publications" class="w3-bar-item w3-button">Research</a>
<a href="#service" class="w3-bar-item w3-button">Services</a>
<a href="#award" class="w3-bar-item w3-button">Awards</a>
</div>
</nav>
<!-- Top menu on small screens -->
<header class="w3-bar w3-top w3-hide-large w3-black w3-xlarge">
<div class="w3-bar-item w3-padding-24">YUNHE</div>
<a href="javascript:void(0)" class="w3-bar-item w3-button w3-padding-24 w3-right" style="font-stretch: extra-expanded;" onclick="w3_open()"><b>≡</b></a>
</div>
</header>
<!-- Overlay effect when opening sidebar on small screens -->
<div class="w3-overlay w3-hide-large" onclick="w3_close()" style="cursor:pointer" title="close side menu" id="myOverlay"></div>
<!-- !PAGE CONTENT! -->
<div class="w3-main" style="margin-left:150px">
<!-- Push down content on small screens -->
<div class="w3-hide-large" style="margin-top:83px"></div>
<!-- The Home Section -->
<div class="w3-container w3-center w3-padding-32" id="home">
<img style="width: 80%;max-width: 320px" alt="profile photo" src="images/Yunhe_new.jpg">
<h1>Yunhe Wang</h1>
<p class="w3-justify" style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;max-width:600px">
I am now the head of the Huawei Applied AI lab and also a senior researcher at <a href="https://www.noahlab.com.hk/">Huawei Noah's Ark Lab</a>, where I work on deep learning, model compression, and computer vision, etc. Before that, I did my PhD at school of EECS, <a href="https://www.pku.edu.cn/">Peking University</a>, where I was co-advised by Prof. <a href="https://dblp.org/pers/hd/x/Xu_0006:Chao">Chao Xu</a></a> and Prof. <a href="https://scholar.google.com.sg/citations?user=RwlJNLcAAAAJ">Dacheng Tao</a></a>. I did my bachelors at school of science, <a href="https://en.xidian.edu.cn/">Xidian University</a>.
</p>
<p class="w3-center">
<a href="mailto:[email protected]">Email</a>  / 
<a href="https://scholar.google.com/citations?user=isizOkYAAAAJ">Google Scholar</a>  / 
<a href="https://www.zhihu.com/people/YunheWang"> Zhi Hu </a>  / 
<a href="https://dblp.org/pid/63/8217-1.html"> DBLP </a>
</p>
</tbody></table>
</div>
<!-- The News Section -->
<div class="w3-container w3-light-grey w3-padding-32" id="news">
<h2>News</h2>
<p><li> <a style="color:red">12/2023, We recently developed PanGu-π: an enhanced LLM architecture via Nonlinearity Compensation</a>.</li></p>
<p><li> 05/2023, I will give a talk about Multimodal Learning at <a href="http://valser.org/2023/#/workshopde?id=11">VALSE 2023 workshop</a>.</li></p>
<p><li> 04/2023, I will give a talk about Robust Machine Learning at <a href="https://conf.ccf.org.cn/web/html4/index.html?globalId=m1104419495071584256168335572674&type=1">The CCF Advanced Disciplines Lectures</a>.</li></p>
<p><li> 03/2023, I accepted the invitation to serve as an Area Chair for <a href="https://nips.cc/Conferences/2023/">NeurIPS 2023</a>.</li></p>
<p><li> 03/2023, 4 papers have been accepted by <a href="http://cvpr2023.theRcvf.com/">CVPR 2023</a>.</li></p>
<p><li> 12/2022, I accepted the invitation to serve as an Area Chair for <a href="https://icml.cc/Conferences/2023">ICML 2023</a>.</li></p>
<p><li> 12/2022, I will give a talk about Efficient Deep Networks at <a href="https://ccf.org.cn/cncc2022/schedule_d_4179">China National Computer Congress (CNCC) 2022</a>.</li></p>
<p><li> 09/2022, 9 papers have been accepted by <a href="https://nips.cc/Conferences/2022">NeurIPS 2022</a>.</li></p>
<p><li> 05/2022, I will give a talk about Vision Transformer at <a href="https://mp.weixin.qq.com/s/wO5RMXDvjPSeiZsIh2xvUw">BAAI 2022</a>.</li></p>
<p><li> 02/2022, 8 papers have been accepted by <a href="http://cvpr2022.theRcvf.com/">CVPR 2022</a>.</li></p>
<p><li> 02/2022, Our suvery paper on vision transformer has been accepted by <a href="https://arxiv.org/pdf/2012.12556">IEEE TPAMI</a>.</li></p>
<!--
<p><li> 09/2021, 10 papers have been accepted by <a href="https://nips.cc/Conferences/2021">NeurIPS 2021</a>. Congrats to everyone!</li></p>
<p><li> 09/2021, The journal version of versatile filters has been accepted by <a href="https://ieeexplore.ieee.org/iel7/34/4359286/09543586.pdf">IEEE TPAMI</a>.</li></p>
<p><li> 07/2021, 1 paper has been accepted by <a href="https://iccv2021.thecvf.com/home">ICCV 2021</a>.</li></p>
<p><li> 05/2021, 1 paper has been accepted by <a href="https://icml.cc/">ICML 2021</a>.</li></p>
<p><li> 05/2021, I have been selected as a Senior Area Chair for <a href="http://valser.org/">VALSE 2021</a>.</li></p>
<p><li> 03/2021, I accepted the invitation to serve as an Area Chair for <a href="https://nips.cc/Conferences/2021/">NeurIPS 2021</a>.</li></p>
<p><li> 03/2021, 9 papers have been accepted by <a href="http://cvpr2021.theRcvf.com/">CVPR 2021</a>.</li></p>
<p><li> 01/2021, I will give a talk about AdderNet at <a href="https://haet2021.github.io/speakers.html">HAET ICLR 2021 workshop</a>.</li></p>
<p><li> 12/2020, two papers have been accepted by <a href="https://aaai.org/Conferences/AAAI-21/">AAAI 2021</a>.</li></p>
<p><li> 11/2020, I accepted the invitation to serve as an Area Chair for <a href="https://icml.cc/Conferences/2021">ICML 2021</a>.</li></p>
<p><li> 09/2020, six papers have been accepted by <a href="https://nips.cc/Conferences/2020">NeurIPS 2020</a>.</li></p>
<p><li> 07/2020, one paper has been accepted by <a href="https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=5962385">IEEE TNNLS</a>.</li></p>
<p><li> 07/2020, one paper has been accepted by <a href="https://eccv2020.eu/accepted-papers/">ECCV 2020</a>.</li></p>
<p><li> 06/2020, two papers have been accepted by <a href="https://icml.cc/Conferences/2020/AcceptedPapersInitial">ICML 2020</a>.</li></p>
<p><li> 07/2020, one paper has been accepted by <a href="http://2020.acmmm.org/accepted-paper-id-list.txt">ACM MM 2020</a>.</li></p>
<p><li> 02/2020, seven papers have been accepted by <a href="http://openaccess.thecvf.com/menu.py">CVPR 2020</a>.</li></p>
<p><li> 01/2020, one paper has been accepted by <a href="https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=5962385">IEEE TNNLS</a>.</li></p>
<p><li> 11/2019, three papers have been accepted by <a href="https://aaai.org/Conferences/AAAI-20/wp-content/uploads/2020/01/AAAI-20-Accepted-Paper-List.pdf">AAAI 2020</a>.</li></p>
-->
</div>
<!-- The Projects Section -->
<div class="w3-container w3-padding-32" id="projects">
<h2>Recent Projects</h2>
<p class="w3-justify">
Actually, model compression is a kind of technique for developing portable deep neural networks with lower memory and computation costs. I have done several projects in Huawei including some smartphones' applications in 2019 and 2020 (e.g. Mate 30 and Honor V30). Currently, I am leading the AdderNet project, which aims to develop a series of deep learning models using only additions (<a href="https://www.reddit.com/r/MachineLearning/comments/ekw2s1/r_addernet_do_we_really_need_multiplications_in/">Discussions on Reddit</a>).
</p>
<h4><li>PanGU-π</li></h4>
<img style="width:96%;" src="images/pangu-pi.PNG">
<p class="w3-justify">
<a style="color: #447ec9" href="data/2023 arxiv pangu-pi.pdf">Paper</a>
</p>
<p class="w3-justify">
Introducing PanGu-π, <span style="color:red">a new architecture for Large Language Model.</span> As the world of Large Language Models (LLMs) continues to evolve with larger models and datasets for enhanced performance, the critical aspect of LLM architecture improvement often remains overlooked. PanGu-π addresses this gap by introducing modules that significantly enhance nonlinearity, thereby greatly boosting the model's expressive capabilities. Achieving leading performance and efficiency in both 7B and 1B model scales, PanGu-π is a testament to the power of architectural innovation in LLMs. Further extending its impact, the specialized YunShan model is making waves in high-value domains such as finance and law, showcasing the practical and powerful application of this groundbreaking technology.
</p>
<h4><li>The Vanilla Neural Architecture for the 2020s</li></h4>
<img style="width:96%;" src="images/VanillaNet.png">
<p class="w3-justify">
<a style="color: #447ec9" href="https://github.com/huawei-noah/VanillaNet">Project Page</a> | <a style="color: #447ec9" href="https://arxiv.org/abs/2305.12972">Paper</a> | <a style="color: #447ec9" href="https://www.zhihu.com/question/531529633/answer/3047230939">Discussion on Zhihu</a>
</p>
<p class="w3-justify">
<span style="color:red">VanillaNet is remarkable!</span> The concept was born from embracing the "less is more" philosophy in computer vision. It's elegantly designed by avoiding intricate depth and operations, such as self-attention, making it remarkably powerful yet concise. The 6-layer VanillaNet surpasses ResNet-34, and the 13-layer variant achieves about 83% Top-1 accuracy, outpacing the performance of networks with hundreds of layers, and revealing exceptional hardware efficiency advantages.
</p>
<h4><li>Adder Neural Networks</li></h4>
<img style="width:96%;" src="images/AdderNet.jpg">
<p class="w3-justify">
<a style="color: #447ec9" href="https://github.com/huawei-noah/AdderNet">Project Page</a> | <a style="color: #447ec9" href="https://arxiv.org/pdf/2101.10015.pdf">Hardware Implementation</a>
</p>
<p class="w3-justify">
I would like to say, <span style="color:red">AdderNet is very cool!</span> The initial idea was came up in about 2017 when climbing with some friends at Beijing. By replacing all convolutional layers (except the first and the last layers), we now can obtain comparable performance on ResNet architectures. In addition, to make the story more complete, we recent release the hardware implementation and some quantization methods. The results are quite encouraging, we can reduce both <strong>the energy consumption and thecircuit areas significantly without affecting the performance</strong>. Now, we are working on more applications to reduce the costs of launching AI algorithms such as low-level vision, detection, and NLP tasks.
</p>
<h4><li>GhostNet on MindSpore: SOTA Lightweight CV Networks</li></h4>
<img style="width:96%;" src="images/GhostNet.png">
<p class="w3-justify">
<a style="color: #447ec9" href="https://live.huawei.com/huaweiconnect/meeting/cn/5872.html">Huawei Connect (HC) 2020</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub">MindSpore Hub</a>
</p>
<p class="w3-justify">
The initial verison of GhostNet was accepted by CVPR 2020, which achieved SOTA performance on ImageNet: <span style="color:red">75.7%</span> top1 acc with only <span style="color:red">226M FLOPS</span>. In the current version, we release a series computer vision models (e.g. int8 quantization, detection, and larger networks) on <strong>MindsSpore 1.0</strong> and <strong>Mate 30 Pro (Kirin 990)</strong>.
</p>
<h4><li>AI on Ascend: Real-Time Video Style Transfer</li></h4>
<img style="width:32%;" src="images/atlas200.png">    <img style="width:60%;" src="images/video.gif">
<p class="w3-justify">
<a style="color: #447ec9" href="https://www.huaweicloud.com/intl/en-us/HDC.Cloud.html">Huawei Developer Conference (HDC) 2020</a> | <a style="color: #447ec9" href="https://developer.huaweicloud.com/exhibition/Atlas_neural_style.html">Online Demo</a>
</p>
<p class="w3-justify">
This project aims to develop a video style transfer system on the <strong>Huawei Atlas 200 DK AI developer Kit</strong>. The latency of the original model for processing one image is about <span style="color:red">630ms</span>. After accelerating it using our method, the lantency now is about <span style="color:red">40ms</span>.
</p>
</div>
<!-- The Talks Section -->
<div class="w3-container w3-light-grey w3-padding-32" id="talks">
<h2>Talks</h2>
<p><li> 12/2022, Hardware Efficient Deep Learning at <a href="https://ccf.org.cn/cncc2022/schedule_d_4179">China National Computer Congress (CNCC) 2022</a>. Thanks Prof. <a href="http://www.nlpr.ia.ac.cn/jcheng/">Jian Cheng</a> for the invitation.</li></p>
<p><li> 05/2022, Low-Level Vision Transformer and Model Compression at <a href="https://2022.baai.ac.cn/">BAAI Conference 2022</a>. Thanks Prof. <a href="https://people.ucas.edu.cn/~sgshan?language=en">Shiguang Shan</a> for the invitation.</li></p>
<p><li> 10/2021, Vision Transformer at <a href="http://valser.org/2021/#/tutorial">VALSE 2021 Tutorial</a>. Thanks Prof. <a href="https://people.ucas.edu.cn/~sgshan?language=en">Shiguang Shan</a> for the invitation.</li></p>
<p><li> 05/2021, Adder Neural Network at <a href="https://haet2021.github.io/speakers.html">HAET ICLR 2021 workshop</a>. Thanks <a href="https://datawisdom.ca/">Vahid Partovi Nia</a> for the invitation.</li></p>
<p><li> 06/2020, "<a href="http://valser.org/webinar/slide/slides/20200603/%E6%A8%A1%E5%9E%8B%E5%8E%8B%E7%BC%A9-%E5%B7%A5%E4%B8%9A%E7%95%8C%E5%92%8C%E5%AD%A6%E6%9C%AF%E7%95%8C%E7%9A%84%E5%B7%AE%E5%BC%82.pdf">AI on the Edge - Discussion on the Gap Between Industry and Academia</a>" at <a href="http://valser.org/">VALSE Webinar.</li>
<p><li> 05/2020, "<a href="https://www.bilibili.com/video/av925692420/">Edge AI: Progress and Future Directions</a>" at <a href="https://www.qbitai.com/">QbitAI</a>.</li></p>
</div>
<!-- The Publications Section -->
<div class="w3-container w3-padding-32"" id="publications">
<h2>Research</h2>
<p class="w3-left-align" style="line-height:200%">
I'm interested in devleoping <strong>efficient models</strong> for computer vision (e.g. classification, detection, and super-resolution) using pruning, quantization, distilaltion, NAS, etc.
</p>
<h4> Preprint Papers:</h4>
<ol>
<p>
<li><strong>PanGu-π: Enhancing Language Model Architectures via Nonlinearity Compensation</strong>
<br>
<strong>Yunhe Wang</strong>, Hanting Chen, Yehui Tang, Tianyu Guo, Kai Han, Ying Nie, Xutao Wang, Hailin Hu, Zheyuan Bai, Yun Wang, Fangcheng Liu, Zhicheng Liu, Jianyuan Guo, Sinan Zeng, Yinchen Zhang, Qinghua Xu, Qun Liu, Jun Yao, Chao Xu, Dacheng Tao
<br>
<em>arXiv</em> 2023.12.xx submitted | <a style="color: #447ec9" href="data/2023 arxiv pangu-pi.pdf">paper</a>
</ol>
<h4> Conference Papers:</h4>
<ol>
<!--
<p>
<li><strong>Title</strong>
<br>
Author, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="paper link">paper</a> | <a style="color: #447ec9" href="https://github.com/">code</a> | <span style="color:red">Oral Presentation</span>
</p>
-->
<p>
<li><strong>Accelerating Sparse Convolution with Column Vector-Wise Sparsity</strong>
<br>
Yijun Tan, Kai Han, Kang Zhao, Xianzhi Yu, Zidong Du, Yunji Chen, <strong>Yunhe Wang</strong>, Jun Yao
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://openreview.net/pdf?id=Q5kXC6hCr1">paper</a>
<p>
<li><strong>Learning Efficient Vision Transformers via Fine-Grained Manifold Distillation</strong>
<br>
Zhiwei Hao, Jianyuan Guo, Ding Jia, Kai Han, Yehui Tang, Chao Zhang, Han Hu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://openreview.net/pdf?id=crFMP5irwzn">paper</a>
</p>
<p>
<li><strong>A Transformer-Based Object Detector with Coarse-Fine Crossing Representations</strong>
<br>
Zhishan Li, Ying Nie, Kai Han, Jianyuan Guo, Lei Xie, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper_files/paper/2022/file/fcfad93e2f30ab4c22f9ec5edfbb5cc0-Paper-Conference.pdf">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/CFDT">MindSpore code</a>
</p>
<p>
<li><strong>Bridge the Gap Between Architecture Spaces via A Cross-Domain Predictor</strong>
<br>
Yuqiao Liu, Yehui Tang, Zeqiong Lv, <strong>Yunhe Wang</strong>, Yanan Sun
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper_files/paper/2022/file/572aaddf9ff774f7c1cf3d0c81c7185b-Paper-Conference.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/lyq998/CDP">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/cdp">MindSpore code</a>
</p>
<p>
<li><strong>Random Normalization Aggregation for Adversarial Defense</strong>
<br>
Minjing Dong, Xinghao Chen, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper_files/paper/2022/file/da3d4d2e9b37f78ec3e7d0428c9b819a-Paper-Conference.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/UniSerj/Random-Norm-Aggregation">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/RNA">MindSpore code</a>
</p>
<p>
<li><strong>Redistribution of Weights and Activations for AdderNet Quantization</strong>
<br>
Ying Nie, Kai Han, Haikang Diao, Chuanjian Liu, Enhua Wu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2212.10200">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/AdderQuant">MindSpore code</a>
</p>
<p>
<li><strong>BiMLP: Compact Binary Architectures for Vision Multi-Layer Perceptrons</strong>
<br>
Yixing Xu, Xinghao Chen, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2212.14158">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/BiMLP">MindSpore code</a> | <span style="color:red">Spotlight</span>
</p>
<p>
<li><strong>GhostNetV2: Enhance Cheap Operation with Long-Range Attention</strong>
<br>
Yehui Tang, Kai Han, Jianyuan Guo, Chang Xu, Chao Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2022 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2211.12905">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Efficient-AI-Backbones/tree/master/ghostnetv2_pytorch">code</a> | <span style="color:red">Spotlight</span>
</p>
<p>
<li><strong>Vision GNN: An Image is Worth Graph of Nodes</strong>
<br>
Kai Han*, <strong>Yunhe Wang</strong>*, Jianyuan Guo, Yehui Tang, Enhua Wu
<br>
<em>NeurIPS</em> 2022 (* equal contribution) | <a style="color: #447ec9" href="https://arxiv.org/pdf/2206.00272.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Efficient-AI-Backbones">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models">MindSpore code</a>
</p>
<p>
<li><strong>Spatial-Channel Token Distillation for Vision MLPs</strong>
<br>
Yanxi Li, Xinghao Chen, Minjing Dong, Yehui Tang, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>ICML</em> 2022 | <a style="color: #447ec9" href="https://proceedings.mlr.press/v162/li22c/li22c.pdf">paper</a>
</p>
<p>
<li><strong>Federated Learning with Positive and Unlabeled Data</strong>
<br>
Xinyang Lin, Hanting Chen, Yixing Xu, Chao Xu, Xiaolin Gui, Yiping Deng, <strong>Yunhe Wang</strong>
<br>
<em>ICML</em> 2022 | <a style="color: #447ec9" href="https://proceedings.mlr.press/v162/lin22b/lin22b.pdf">paper</a>
</p>
<p>
<li><strong>Brain-inspired Multilayer Perceptron with Spiking Neurons</strong>
<br>
Wenshuo Li, Hanting Chen, Jianyuan Guo, Ziyang Zhang, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Li_Brain-Inspired_Multilayer_Perceptron_With_Spiking_Neurons_CVPR_2022_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/snn-mlp">MindSpore code</a>
</p>
<p>
<li><strong>Source-Free Domain Adaptation via Distribution Estimation</strong>
<br>
Ning Ding, Yixing Xu, Yehui Tang, Chao Xu, <strong>Yunhe Wang</strong>, Dacheng Tao
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Ding_Source-Free_Domain_Adaptation_via_Distribution_Estimation_CVPR_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>Multimodal Token Fusion for Vision Transformers</strong>
<br>
Yikai Wang, Xinghao Chen, Lele Cao, Wenbing Huang, Fuchun Sun, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Wang_Multimodal_Token_Fusion_for_Vision_Transformers_CVPR_2022_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/noah-research">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/TokenFusion">MindSpore code</a>
</p>
<p>
<li><strong>An Image Patch is a Wave: Phase-Aware Vision MLP</strong>
<br>
Yehui Tang, Kai Han, Jianyuan Guo, Chang Xu, Yanxi Li, Chao Xu, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Tang_An_Image_Patch_Is_a_Wave_Phase-Aware_Vision_MLP_CVPR_2022_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/CV-Backbones/tree/master/wavemlp_pytorch">code</a> | <span style="color:red">Oral Presentation</span>
</p>
<p>
<li><strong>Instance-Aware Dynamic Neural Network Quantization</strong>
<br>
Zhenhua Liu, <strong>Yunhe Wang</strong>, Kai Han, Siwei Ma, Wen Gao
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Liu_Instance-Aware_Dynamic_Neural_Network_Quantization_CVPR_2022_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Efficient-Computing">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/DynamicQuant">MindSpore code</a> | <span style="color:red">Oral Presentation</span>
</p>
<p>
<li><strong>Hire-MLP: Vision MLP via Hierarchical Rearrangement</strong>
<br>
Jianyuan Guo, Yehui Tang, Kai Han, Xinghao Chen, Han Wu, Chao Xu, Chang Xu, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Guo_Hire-MLP_Vision_MLP_via_Hierarchical_Rearrangement_CVPR_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>CMT: Convolutional Neural Networks Meet Vision Transformers</strong>
<br>
Jianyuan Guo, Kai Han, Han Wu, Yehui Tang, Xinghao Chen, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Guo_CMT_Convolutional_Neural_Networks_Meet_Vision_Transformers_CVPR_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>Patch Slimming for Efficient Vision Transformers</strong>
<br>
Yehui Tang, Kai Han, <strong>Yunhe Wang</strong>, Chang Xu, Jianyuan Guo, Chao Xu, Dacheng Tao
<br>
<em>CVPR</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022/papers/Tang_Patch_Slimming_for_Efficient_Vision_Transformers_CVPR_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>Transformer in Transformer</strong>
<br>
Kai Han, An Xiao, Enhua Wu, Jianyuan Guo, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/854d9fca60b4bd07f9bb215d59ef5561-Paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/CV-Backbones">code</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/models/tree/master/research/cv/TNT">MindSpore code</a>
</p>
<p>
<li><strong>Learning Frequency Domain Approximation for Binary Neural Networks</strong>
<br>
Yixing Xu, Kai Han, Chang Xu, Yehui Tang, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/d645920e395fedad7bbbed0eca3fe2e0-Paper.pdf">paper</a> | <span style="color:red">Oral Presentation</span>
</p>
<p>
<li><strong>Dynamic Resolution Network</strong>
<br>
Mingjian Zhu*, Kai Han*, Enhua Wu, Qiulin Zhang, Ying Nie, Zhenzhong Lan, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 (* equal contribution) | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/e56954b4f6347e897f954495eab16a88-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Post-Training Quantization for Vision Transformer</strong>
<br>
Zhenhua Liu, <strong>Yunhe Wang</strong>, Kai Han, Wei Zhang, Siwei Ma, Wen Gao
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/ec8956637a99787bd197eacd77acce5e-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Augmented Shortcuts for Vision Transformers</strong>
<br>
Yehui Tang, Kai Han, Chang Xu, An Xiao, Yiping Deng, Chao Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/818f4654ed39a1c147d1e51a00ffb4cb-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Adder Attention for Vision Transformer</strong>
<br>
Han Shu*, Jiahao Wang*, Hanting Chen, Lin Li, Yujiu Yang, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 (* equal contribution) | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/a57e8915461b83adefb011530b711704-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Towards Stable and Robust Addernets</strong>
<br>
Minjing Dong, <strong>Yunhe Wang</strong>, Xinghao Chen, Chang Xu
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/6e3197aae95c2ff8fcab35cb730f6a86-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Handling Long-Tailed Feature Distribution in Addernets</strong>
<br>
Minjing Dong, <strong>Yunhe Wang</strong>, Xinghao Chen, Chang Xu
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/95323660ed2124450caaac2c46b5ed90-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Neural Architecture Dilation for Adversarial Robustness</strong>
<br>
Yanxi Li, Zhaohui Yang, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/f7664060cc52bc6f3d620bcedc94a4b6-Paper.pdf">paper</a>
</p>
<p>
<li><strong>An Empirical Study of Adder Neural Networks for Object Detection</strong>
<br>
Xinghao Chen, Chang Xu, Minjing Dong, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2021 | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2021/file/37693cfc748049e45d87b8c7d8b9aacd-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Learning Frequency-Aware Dynamic Network for Efficient Super-Resolution</strong>
<br>
Wenbin Xie, Dehua Song, Chang Xu, Chunjing Xu, Hui Zhang, <strong>Yunhe Wang</strong>
<br>
<em>ICCV</em> 2021 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content/ICCV2021/papers/Xie_Learning_Frequency-Aware_Dynamic_Network_for_Efficient_Super-Resolution_ICCV_2021_paper.pdf">paper</a>
</p>
<p>
<li><strong>Winograd Algorithm for AdderNet</strong>
<br>
Wenshuo Li, Hanting Chen, Mingqiang Huang, Xinghao Chen, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>ICML</em> 2021 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2105.05530.pdf">paper</a>
</p>
<p>
<li><strong>Distilling Object Detectors via Decoupled Features</strong>
<br>
Jianyuan Guo, Kai Han, <strong>Yunhe Wang</strong>, Wei Zhang, Chunjing Xu, Chang Xu
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2021/papers/Guo_Distilling_Object_Detectors_via_Decoupled_Features_CVPR_2021_paper.pdf">paper</a>
</p>
<p>
<li><strong>HourNAS: Extremely Fast Neural Architecture Search Through an Hourglass Lens</strong>
<br>
Zhaohui Yang, <strong>Yunhe Wang</strong>, Xinghao Chen, Jianyuan Guo, Wei Zhang,
<br>
Chao Xu, Chunjing Xu, Dacheng Tao, Chang Xu
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2005.14446.pdf">paper</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub/details?noah-cvlab/gpu/1.1/HourNAS-F_v1.0_cifar10">MindSpore code</a>
</p>
<p>
<li><strong>Manifold Regularized Dynamic Network Pruning</strong>
<br>
Yehui Tang, <strong>Yunhe Wang</strong>, Yixing Xu, Yiping Deng, Chao Xu, Dacheng Tao, Chang Xu
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2103.05861.pdf">paper</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub/details?noah-cvlab/gpu/1.1/manidp_v1.0_cifar10">MindSpore code</a>
</p>
<p>
<li><strong>Learning Student Networks in the Wild</strong>
<br>
Hanting Chen, Tianyu Guo, Chang Xu, Wenshuo Li, Chunjing Xu, Chao Xu, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2021/papers/Chen_Learning_Student_Networks_in_the_Wild_CVPR_2021_paper.pdf">paper</a>
</p>
<p>
<li><strong>AdderSR: Towards Energy Efficient Image Super-Resolution</strong>
<br>
Dehua Song*, <strong>Yunhe Wang</strong>*, Hanting Chen, Chang Xu, Chunjing Xu, Dacheng Tao
<br>
<em>CVPR</em> 2021 (* equal contribution) | <a style="color: #447ec9" href="https://arxiv.org/pdf/2009.08891.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/AdderNet">code</a> | <span style="color:red">Oral Presentation</span>
</p>
<p>
<li><strong>ReNAS: Relativistic Evaluation of Neural Architecture Search</strong>
<br>
Yixing Xu, <strong>Yunhe Wang</strong>, Kai Han, Yehui Tang, Shangling Jui, Chunjing Xu, Chang Xu
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://arxiv.org/pdf/1910.01523.pdf">paper</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub/details?noah-cvlab/gpu/1.1/renas_v1.0_cifar10">MindSpore code</a> | <span style="color:red">Oral Presentation</span>
</p>
<p>
<li><strong>Pre-Trained Image Processing Transformer</strong>
<br>
Hanting Chen, <strong>Yunhe Wang</strong>, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu,
<br>
Siwei Ma, Chunjing Xu, Chao Xu, Wen Gao
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2012.00364.pdf">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/IPT">MindSpore code</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Pretrained-IPT">Pytorch code</a>
</p>
<p>
<li><strong>Data-Free Knowledge Distillation For Image Super-Resolution</strong>
<br>
Yiman Zhang, Hanting Chen, Xinghao Chen, Yiping Deng, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content/CVPR2021/papers/Zhang_Data-Free_Knowledge_Distillation_for_Image_Super-Resolution_CVPR_2021_paper.pdf">paper</a>
</p>
<p>
<li><strong>Positive-Unlabeled Data Purification in the Wild for Object Detection</strong>
<br>
Jianyuan Guo, Kai Han, Han Wu, Xinghao Chen, Chao Zhang, Chunjing Xu, Chang Xu, <strong>Yunhe Wang</strong>
<br>
<em>CVPR</em> 2021 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content/CVPR2021/papers/Guo_Positive-Unlabeled_Data_Purification_in_the_Wild_for_Object_Detection_CVPR_2021_paper.pdf">paper</a>
</p>
<p>
<li><strong>One-shot Graph Neural Architecture Search with Dynamic Search Space</strong>
<br>
Yanxi Li, Zean Wen, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>AAAI</em> 2021 <a style="color: #447ec9" href="https://www.aaai.org/AAAI21Papers/AAAI-3955.YangS.pdf">paper</a>
</p>
<p>
<li><strong>Adversarial Robustness through Disentangled Representations</strong>
<br>
Shuo Yang, Tianyu Guo, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>AAAI</em> 2021 <a style="color: #447ec9" href="https://www.aaai.org/AAAI21Papers/AAAI-3955.YangS.pdf">paper</a>
</p>
<p>
<li><strong>Kernel Based Progressive Distillation for Adder Neural Networks</strong>
<br>
Yixing Xu, Chang Xu, Xinghao Chen, Wei Zhang, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>NeurIPS</em> 2020 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2009.13044.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/AdderNet">code</a> | <span style="color:red"> Spotlight</span>
</p>
<p>
<li><strong>Model Rubik's Cube: Twisting Resolution, Depth and Width for TinyNets</strong>
<br>
Kai Han*, <strong>Yunhe Wang</strong>*, Qiulin Zhang, Wei Zhang, Chunjing Xu, Tong Zhang
<br>
<em>NeurIPS</em> 2020 (* equal contribution) | <a style="color: #447ec9" href="https://arxiv.org/pdf/2010.14819.pdf">paper</a> | <a style="color: #447ec9" href="https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/tinynet">code</a>
</p>
<p>
<li><strong>Residual Distillation: Towards Portable Deep Neural Networks without Shortcuts</strong>
<br>
Guilin Li*, Junlei Zhang*, <strong>Yunhe Wang</strong>, Chuanjian Liu, Matthias Tan, Yunfeng Lin,
<br>
Wei Zhang, Jiashi Feng, Tong Zhang
<br>
<em>NeurIPS</em> 2020 (* equal contribution) | <a style="color: #447ec9" href="https://proceedings.neurips.cc/paper/2020/file/657b96f0592803e25a4f07166fff289a-Paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/leoozy/JointRD_Neurips2020">code</a>
</p>
<p>
<li><strong>Searching for Low-Bit Weights in Quantized Neural Networks</strong>
<br>
Zhaohui Yang, <strong>Yunhe Wang</strong>, Kai Han, Chunjing Xu, Chao Xu, Dacheng Tao, Chang Xu
<br>
<em>NeurIPS</em> 2020 | <a style="color: #447ec9" href="https://arxiv.org/abs/2009.08695.pdf">paper</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub/details?noah-cvlab/gpu/1.0/VGG-Small-low-bit_cifar10">code</a>
</p>
<p>
<li><strong>SCOP: Scientific Control for Reliable Neural Network Pruning</strong>
<br>
Yehui Tang, <strong>Yunhe Wang</strong>, Yixing Xu, Dacheng Tao, Chunjing Xu, Chao Xu, Chang Xu
<br>
<em>NeurIPS</em> 2020 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2010.10732">paper</a> | <a style="color: #447ec9" href="https://www.mindspore.cn/resources/hub/details?2593/noah-cvlab/gpu/1.0/resnet-0.65x_v1.0_oxford_pets">code</a>
</p>
<p>
<li><strong>Adapting Neural Architectures Between Domains</strong>
<br>
Yanxi Li, Zhaohui Yang, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>NeurIPS</em> 2020 | <a style="color: #447ec9" href="https://papers.nips.cc/paper/2020/file/08f38e0434442128fab5ead6217ca759-Paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/liyxi/AdaptNAS">code</a>
</p>
<p>
<li><strong>Discernible Image Compression</strong>
<br>
Zhaohui Yang, <strong>Yunhe Wang</strong>, Chang Xu, Peng Du, Chao Xu, Chunjing Xu, Qi Tian
<br>
<em>ACM MM</em> 2020 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2002.06810">paper</a>
</p>
<p>
<li><strong>Optical Flow Distillation: Towards Efficient and Stable Video Style Transfer</strong>
<br>
Xinghao Chen*, Yiman Zhang*, <strong>Yunhe Wang</strong>, Han Shu, Chunjing Xu, Chang Xu
<br>
<em>ECCV</em> 2020 (* equal contribution) | <a style="color: #447ec9" href="https://arxiv.org/pdf/2007.05146.pdf">paper</a> | <a style="color: #447ec9" href="https://gitee.com/AtlasCase/sample-styletransfer">code</a>
</p>
<p>
<li><strong>Learning Binary Neurons with Noisy Supervision</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Yixing Xu, Chunjing Xu, Enhua Wu, Chang Xu
<br>
<em>ICML</em> 2020 | <a style="color: #447ec9" href="https://proceedings.icml.cc/static/paper_files/icml/2020/181-Paper.pdf">paper</a>
</p>
<p>
<li><strong>Neural Architecture Search in a Proxy Validation Loss Landscape</strong>
<br>
Yanxi Li, Minjing Dong, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>ICML</em> 2020 | <a style="color: #447ec9" href="https://proceedings.icml.cc/static/paper_files/icml/2020/439-Paper.pdf">paper</a>
<p>
<li><strong>On Positive-Unlabeled Classification in GAN</strong>
<br>
Tianyu Guo, Chang Xu, Jiajun Huang, <strong>Yunhe Wang</strong>, Boxin Shi, Chao Xu, Dacheng Tao
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Guo_On_Positive-Unlabeled_Classification_in_GAN_CVPR_2020_paper.pdf">paper</a>
</p>
<p>
<li><strong>CARS: Continuous Evolution for Efficient Neural Architecture Search</strong>
<br>
Zhaohui Yang, <strong>Yunhe Wang</strong>, Xinghao Chen, Boxin Shi, Chao Xu, Chunjing Xu, Qi Tian, Chang Xu
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Yang_CARS_Continuous_Evolution_for_Efficient_Neural_Architecture_Search_CVPR_2020_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/CARS">code</a>
</p>
<p>
<li><strong>AdderNet: Do We Really Need Multiplications in Deep Learning?</strong>
<br>
Hanting Chen*, <strong>Yunhe Wang</strong>*, Chunjing Xu, Boxin Shi, Chao Xu, Qi Tian, Chang Xu
<br>
<em>CVPR</em> 2020 (* equal contribution) | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Chen_AdderNet_Do_We_Really_Need_Multiplications_in_Deep_Learning_CVPR_2020_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/AdderNet">code</a> | <span style="color:red">Oral Presentation</span>
<br>
</p>
<p>
<li><strong>A Semi-Supervised Assessor of Neural Architectures</strong>
<br>
Yehui Tang, <strong>Yunhe Wang</strong>, Yixing Xu, Hanting Chen, Boxin Shi, Chao Xu, Chunjing Xu, Qi Tian, Chang Xu
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Tang_A_Semi-Supervised_Assessor_of_Neural_Architectures_CVPR_2020_paper.pdf">paper</a>
</p>
<p>
<li><strong>Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection</strong>
<br>
Jianyuan Guo, Kai Han, <strong>Yunhe Wang</strong>, Chao Zhang, Zhaohui Yang, Han Wu, Xinghao Chen, Chang Xu
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Guo_Hit-Detector_Hierarchical_Trinity_Architecture_Search_for_Object_Detection_CVPR_2020_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/ggjy/HitDet.pytorch">code</a>
</p>
<p>
<li><strong>Frequency Domain Compact 3D Convolutional Neural Networks</strong>
<br>
Hanting Chen, <strong>Yunhe Wang</strong>, Han Shu, Yehui Tang, Chunjing Xu, Boxin Shi, Chao Xu, Qi Tian, Chang Xu
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="http://openaccess.thecvf.com/content_CVPR_2020/papers/Chen_Frequency_Domain_Compact_3D_Convolutional_Neural_Networks_CVPR_2020_paper.pdf">paper</a>
</p>
<li><strong>GhostNet: More Features from Cheap Operations</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Qi Tian, Jianyuan Guo, Chunjing Xu, Chang Xu
<br>
<em>CVPR</em> 2020 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content_CVPR_2020/papers/Han_GhostNet_More_Features_From_Cheap_Operations_CVPR_2020_paper.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/ghostnet">code</a>
</p>
<p>
<li><strong>Beyond Dropout: Feature Map Distortion to Regularize Deep Neural Networks</strong>
<br>
Yehui Tang, <strong>Yunhe Wang</strong>, Yixing Xu, Boxin Shi, Chao Xu, Chunjing Xu, Chang Xu
<br>
<em>AAAI</em> 2020 | <a style="color: #447ec9" href="data/2020 AAAI dropblock.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/disout">code</a>
</p>
<p>
<li><strong>DropNAS: Grouped Operation Dropout for Differentiable Architecture Search</strong>
<br>
Weijun Hong, Guilin Li, Weinan Zhang, Ruiming Tang, <strong>Yunhe Wang</strong>, Zhenguo Li, Yong Yu
<br>
<em>IJCAI</em> 2020 | <a style="color: #447ec9" href="https://www.ijcai.org/Proceedings/2020/0322.pdf">paper</a>
</p>
<p>
<li><strong>Distilling Portable Generative Adversarial Networks for Image Translation</strong>
<br>
Hanting Chen, <strong>Yunhe Wang</strong>, Han Shu, Changyuan Wen, Chunjing Xu, Boxin Shi, Chao Xu, Chang Xu
<br>
<em>AAAI</em> 2020 | <a style="color: #447ec9" href="data/2020 AAAI GAN Distillation.pdf">paper</a>
</p>
<p>
<li><strong>Efficient Residual Dense Block Search for Image Super-Resolution</strong>
<br>
Dehua Song, Chang Xu, Xu Jia, Yiyi Chen, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>AAAI</em>, 2020 | <a style="color: #447ec9" href="data/2020 AAAI SR NAS.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/vega">code</a>
</p>
<p>
<li><strong>Positive-Unlabeled Compression on the Cloud</strong>
<br>
Yixing Xu, <strong>Yunhe Wang</strong>, Hanting Chen, Kai Han, Chunjing Xu, Dacheng Tao, Chang Xu
<br>
<em>NeurIPS</em> 2019 | <a style="color: #447ec9" href="data/2019 NIPS PU Compression.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Data-Efficient-Model-Compression/tree/master/pu_compress">code</a> | <a style="color: #447ec9" href="http://papers.nips.cc/paper/8525-positive-unlabeled-compression-on-the-cloud-supplemental.zip">supplement</a>
</p>
<p>
<li><strong>Data-Free Learning of Student Networks</strong>
<br>
Hanting Chen,<strong>Yunhe Wang</strong>, Chang Xu, Zhaohui Yang, Chuanjian Liu, Boxin Shi,
<br>
Chunjing Xu, Chao Xu, Qi Tian
<br>
<em>ICCV</em> 2019 | <a style="color: #447ec9" href="data/2019 ICCV DAFL.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Data-Efficient-Model-Compression/tree/master/DAFL">code</a>
</p>
<p>
<li><strong>Co-Evolutionary Compression for Unpaired Image Translation</strong>
<br>
Han Shu, <strong>Yunhe Wang</strong>, Xu Jia, Kai Han, Hanting Chen, Chunjing Xu, Qi Tian, Chang Xu
<br>
<em>ICCV</em> 2019 | <a style="color: #447ec9" href="data/2019 ICCV Co-Evolution Pruning.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/GAN-pruning">code</a>
</p>
<p>
<li><strong>LegoNet: Efficient Convolutional Neural Networks with Lego Filters</strong>
<br>
Zhaohui Yang, <strong>Yunhe Wang</strong>, Hanting Chen, Chuanjian Liu, Boxin Shi, Chao Xu, Chunjing Xu, Chang Xu
<br>
<em>ICML</em> 2019 | <a style="color: #447ec9" href="data/2019 ICML LegoNet.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/LegoNet">code</a>
</p>
<p>
<li><strong>Learning Instance-wise Sparsity for Accelerating Deep Models</strong>
<br>
Chuanjian Liu, <strong>Yunhe Wang</strong>, Kai Han, Chunjing Xu, Chang Xu
<br>
<em>IJCAI</em> 2019 | <a style="color: #447ec9" href="data/2019 IJCAI Instance Sparsity.pdf">paper</a>
</p>
<p>
<li><strong>Attribute Aware Pooling for Pedestrian Attribute Recognition</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Han Shu, Chuanjian Liu, Chunjing Xu, Chang Xu
<br>
<em>IJCAI</em> 2019 | <a style="color: #447ec9" href="https://www.ijcai.org/proceedings/2019/0341.pdf">paper</a>
</p>
<p>
<li><strong>Crafting Efficient Neural Graph of Large Entropy</strong>
<br>
Minjing Dong, Hanting Chen, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>IJCAI</em> 2019 | <a style="color: #447ec9" href="data/2019 IJCAI Graph Pruning.pdf">paper</a>
</p>
<p>
<li><strong>Low Resolution Visual Recognition via Deep Feature Distillation</strong>
<br>
Mingjian Zhu, Kai Han, Chao Zhang, Jinlong Lin, <strong>Yunhe Wang</strong>
<br>
<em>ICASSP</em> 2019 | <a style="color: #447ec9" href="data/2019 ICASSP LR Distillation.pdf">paper</a>
</p>
<p>
<li><strong>Learning Versatile Filters for Efficient Convolutional Neural Networks</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Chunjing Xu, Chao Xu, Dacheng Tao
<br>
<em>NeurIPS</em> 2018 | <a style="color: #447ec9" href="data/2018 NIPS Versatile Filter.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Versatile-Filters">code</a> | <a style="color: #447ec9" href="https://papers.nips.cc/paper/7433-learning-versatile-filters-for-efficient-convolutional-neural-networks-supplemental.zip">supplement</a>
</p>
<p>
<li><strong>Towards Evolutionary Compression</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Jiayan Qiu, Chao Xu, Dacheng Tao
<br>
<em>SIGKDD</em> 2018 | <a style="color: #447ec9" href="data/2018 KDD GA pruning.pdf">paper</a>
</p>
<p>
<li><strong>Autoencoder Inspired Unsupervised Feature Selection</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Chao Zhang, Chao Li, Chao Xu
<br>
<em>ICASSP</em> 2018 | <a style="color: #447ec9" href="data/2018 ICASSP Feature Selector.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/NoahLuffy/AEFS">code</a>
</p>
<p>
<li><strong>Adversarial Learning of Portable Student Networks</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Chao Xu, Dacheng Tao
<br>
<em>AAAI</em> 2018 | <a style="color: #447ec9" href="data/2018 AAAI Adversarial Distillation.pdf">paper</a>
</p>
<p>
<li><strong>Beyond Filters: Compact Feature Map for Portable Deep Model</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Chao Xu, Dacheng Tao
<br>
<em>ICML</em> 2017 | <a style="color: #447ec9" href="data/2017 ICML Beyond Filters.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/YunheWang/RedCNN">code</a> | <a style="color: #447ec9" href="http://proceedings.mlr.press/v70/wang17m/wang17m-supp.zip">supplement</a>
</p>
<p>
<li><strong>Beyond RPCA: Flattening Complex Noise in the Frequency Domain</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Chao Xu, Dacheng Tao
<br>
<em>AAAI</em> 2017 | <a style="color: #447ec9" href="data/2017 AAAI Beyond RPCA.pdf">paper</a>
</p>
<p>
<li><strong>Privileged Multi-Label Learning</strong>
<br>
Shan You, Chang Xu, <strong>Yunhe Wang</strong>, Chao Xu, Dacheng Tao
<br>
<em>IJCAI</em> 2017 | <a style="color: #447ec9" href="data/2017 IJCAI Privileged.pdf">paper</a>
</p>
<p>
<li><strong>CNNpack: Packing Convolutional Neural Networks in the Frequency Domain</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Shan You, Chao Xu, Dacheng Tao
<br>
<em>NeurIPS</em> 2016 | <a style="color: #447ec9" href="data/2016 NIPS CNNpack.pdf">paper</a> | <a style="color: #447ec9" href="https://papers.nips.cc/paper/6390-cnnpack-packing-convolutional-neural-networks-in-the-frequency-domain-supplemental.zip">supplement</a>
</p>
</ol>
<h4> Journal Papers:</h4>
<ol>
<p>
<li><strong>Neural Architecture Search via Proxy Validation</strong>
<br>
Yanxi Li, Minjing Dong, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>IEEE TPAMI</em> 2022 | <a style="color: #447ec9" href="https://ieeexplore.ieee.org/iel7/34/4359286/09931480.pdf">paper</a>
</p>
<p>
<li><strong>Local Means Binary Networks for Image Super-Resolution</strong>
<br>
Keyu Li, Nannan Wang, Jingwei Xin, Xinrui Jiang, Jie Li, Xinbo Gao, Kai Han, <strong>Yunhe Wang</strong>
<br>
<em>IEEE TNNLS</em> 2022 | <a style="color: #447ec9" href="https://ieeexplore.ieee.org/iel7/5962385/6104215/09927310.pdf">paper</a>
</p>
<p>
<li><strong>GhostNets on Heterogeneous Devices via Cheap Operations</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Chang Xu, Jianyuan Guo, Chunjing Xu, Enhua Wu, Qi Tian
<br>
<em>IJCV</em> 2022 | <a style="color: #447ec9" href="https://link.springer.com/content/pdf/10.1007/s11263-022-01575-y.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Efficient-AI-Backbones">code</a>
</p>
<p>
<li><strong>A Survey on Visual Transformer</strong>
<br>
Kai Han, <strong>Yunhe Wang</strong>, Hanting Chen, Xinghao Chen, Jianyuan Guo, Zhenhua Liu, Yehui Tang, An Xiao, Chunjing Xu, Yixing Xu, Zhaohui Yang, Yiman Zhang, Dacheng Tao
<br>
<em>IEEE TPAMI</em> 2022 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2012.12556">paper</a>
</p>
<p>
<li><strong>Learning Versatile Convolution Filters for Efficient Visual Recognition</strong>
<br>
Kai Han*, <strong>Yunhe Wang</strong>*, Chang Xu, Chunjing Xu, Enhua Wu, Dacheng Tao
<br>
<em>IEEE TPAMI</em> 2021 (* equal contribution) | <a style="color: #447ec9" href="https://ieeexplore.ieee.org/iel7/34/4359286/09543586.pdf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Versatile-Filters">code</a>
</p>
<p>
<li><strong>Adversarial Recurrent Time Series Imputation</strong>
<br>
Shuo Yang, Minjing Dong, <strong>Yunhe Wang</strong>, Chang Xu
<br>
<em>IEEE TNNLS</em> 2020 |<a style="color: #447ec9" href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9158560">paper</a>
</p>
<p>
<li><strong>Learning Student Networks via Feature Embedding</strong>
<br>
Hanting Chen, <strong>Yunhe Wang</strong>, Chang Xu, Chao Xu, Dacheng Tao
<br>
<em>IEEE TNNLS</em> 2020 | <a style="color: #447ec9" href="https://arxiv.org/pdf/1812.06597">paper</a>
</p>
<p>
<li><strong>Packing Convolutional Neural Networks in the Frequency Domain</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Chao Xu, Dacheng Tao
<br>
<em>IEEE TPAMI</em> 2018 | <a style="color: #447ec9" href="data/2018 PAMI CNNpack.pdf">paper</a>
</p>
<p>
<li><strong>DCT Regularized Extreme Visual Recovery</strong>
<br>
<strong>Yunhe Wang</strong>, Chang Xu, Shan You, Chao Xu, Dacheng Tao
<br>
<em>IEEE TIP</em> 2017 | <a style="color: #447ec9" href="data/2017 TIP DCT norm.pdf">paper</a>
</p>
<p>
<li><strong>DCT Inspired Feature Transform for Image Retrieval and Reconstruction</strong>
<br>
<strong>Yunhe Wang</strong>, Miaojing Shi, Shan You, Chao Xu
<br>
<em>IEEE TIP</em> 2016 | <a style="color: #447ec9" href="data/2016 TIP DCT feature.pdf">paper</a>
</p>
</ol>
<h4> Workshop Papers:</h4>
<ol>
<p>
<li><strong>PyramidTNT: Improved Transformer-in-Transformer Baselines with Pyramid Architecture</strong>
<br>
Kai Han, Jianyuan Guo, Yehui Tang, <strong>Yunhe Wang</strong>
<br>
<em>CVPR Workshop</em> 2022 | <a style="color: #447ec9" href="https://arxiv.org/pdf/2201.00978&ved=2ahUKEwj647mwk6D1AhVQzqQKHVadCLMQFnoECBcQAQ&usg=AOvVaw2Cf1MOSBpeApetybQC_Zgf">paper</a> | <a style="color: #447ec9" href="https://github.com/huawei-noah/Efficient-AI-Backbones/tree/master/tnt_pytorch">code</a>
</p>
<p>
<li><strong>Network Amplification with Efficient MACs Allocation</strong>
<br>
Chuanjian Liu, Kai Han, An Xiao, Ying Nie, Wei Zhang, <strong>Yunhe Wang</strong>
<br>
<em>CVPR Workshop</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022W/NAS/papers/Liu_Network_Amplification_With_Efficient_MACs_Allocation_CVPRW_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>Searching for Energy-Efficient Hybrid Adder-Convolution Neural Networks</strong>
<br>
Wenshuo Li, Xinghao Chen, Jinyu Bai, Xuefei Ning, <strong>Yunhe Wang</strong>
<br>
<em>CVPR Workshop</em> 2022 | <a style="color: #447ec9" href="https://openaccess.thecvf.com/content/CVPR2022W/NAS/papers/Li_Searching_for_Energy-Efficient_Hybrid_Adder-Convolution_Neural_Networks_CVPRW_2022_paper.pdf">paper</a>
</p>
<p>
<li><strong>Searching for Accurate Binary Neural Architectures</strong>
<br>
Mingzhu Shen, Kai Han, Chunjing Xu, <strong>Yunhe Wang</strong>
<br>
<em>ICCV Neural Architectures Workshop</em> 2019 | <a style="color: #447ec9" href="data/2019 ICCVw Bianry Search.pdf">paper</a>
</p>
</ol>