-
Notifications
You must be signed in to change notification settings - Fork 0
/
backup_index.html
executable file
·503 lines (470 loc) · 28.3 KB
/
backup_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
<!DOCTYPE HTML>
<html lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<title>Qiyao Wei</title>
<meta name="author" content="Qiyao Wei">
<meta name="viewport" content="width=device-width, initial-scale=1">
<link rel="stylesheet" type="text/css" href="stylesheet.css">
<!-- <link rel="icon" type="image/png" href="images/seal_icon.png">-->
</head>
<body>
<table style="width:100%;max-width:800px;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr style="padding:0px">
<td style="padding:0px">
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr style="padding:0px">
<td style="padding:2.5%;width:63%;vertical-align:middle">
<p style="text-align:center">
<name>Qiyao Wei</name>
</p>
<!-- <a href="https://ai.google/research">Google Research</a>-->
<p>
I am a senior year undergrad at University of Toronto.
Currently, my research interest is using theory to justify our choices in network design and
hyperparameter tuning.
My long-term goal is to crack the code of human intelligence, or to achieve human-level
intelligence in machines.
</p>
<p style="text-align:center">
<a href="mailto:[email protected]">Email</a>  / 
<a href="data/misc/weiqiyaoCV.pdf">CV</a>  / 
<a href="https://github.com/QiyaoWei">Github</a>
</p>
</td>
<td style="padding:2.5%;width:40%;max-width:40%">
<a href="images/me.jpg"><img style="width:100%;max-width:100%" alt="profile photo"
src="images/me.jpg" class="hoverZoomLink"></a>
</td>
</tr>
</tbody>
</table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>
<tr onmouseout="deq_stop()" onmouseover="deq_start()">
<td style="padding:20px;width:25%;vertical-align:middle">
<div class="one">
<div class="two" id='deq_image'>
<video width=100% height=100% muted autoplay loop>
<source src="data/pub/deq/evolution.mp4" type="video/mp4">
Your browser does not support the video tag.
</video>
</div>
<img src='data/pub/deq/final_features.png' width="160">
</div>
<script type="text/javascript">
function deq_start() {
document.getElementById('deq_image').style.opacity = "1";
}
function deq_stop() {
document.getElementById('deq_image').style.opacity = "0";
}
deq_stop()
</script>
</td>
<td style="padding:20px;width:75%;vertical-align:middle">
<a href="https://julialang.org/blog/2021/10/DEQ/">
<papertitle>Composability in Julia: Implementing Deep Equilibrium Models via Neural ODEs</papertitle>
</a>
<br>
<strong>Qiyao Wei</strong>,
<a href="https://frankschae.github.io/">Frank Schäfer</a>,
<a href="https://avik-pal.github.io/">Avik Pal</a>,
<a href="https://chrisrackauckas.com/">Chris Rackauckas</a>,
<br>
<em>Julialang</em>, 2021
<br>
<a href="https://julialang.org/blog/2021/10/DEQ/">Blog</a>
/
<a href="https://julialang.org/blog/2021/10/DEQ/">Code</a>
/
<a href="https://news.ycombinator.com/item?id=28945191">News</a>
<!-- <a href="https://arxiv.org/abs/2103.14645">arXiv</a>-->
<!-- /-->
<!-- <a href="https://www.youtube.com/watch?v=5jKry8n5YO8">video</a>-->
<!-- /-->
<!-- <a href="https://nerf.live/#demos">demo</a>-->
<p></p>
<p>Taking advantage of composability in Julia, we reformulate DEQs using ideas from NeuralODE, leading
to a flexible implementation that could easily be extended.</p>
</td>
</tr>
<tr onmouseout="cmc_stop()" onmouseover="cmc_start()">
<td style="padding:20px;width:25%;vertical-align:middle">
<div class="one">
<div class="two" id='cmc_image'>
<video width=100% height=100% muted autoplay loop>
<source src="data/misc/cmc.png" type="video/mp4">
Your browser does not support the video tag.
</video>
</div>
<img src='data/misc/cmc.png' width="160">
</div>
<script type="text/javascript">
function cmc_start() {
document.getElementById('cmc_image').style.opacity = "1";
}
function cmc_stop() {
document.getElementById('cmc_image').style.opacity = "0";
}
cmc_stop()
</script>
</td>
<td style="padding:20px;width:75%;vertical-align:middle">
<a href="https://github.com/QiyaoWei/Distributed-CMC">
<papertitle>Distributed CMC</papertitle>
</a>
<br>
<strong>Qiyao Wei</strong>,
<a href="https://ai.stanford.edu/~optas/">Panos Achlioptas</a>,
<br>
<em>Github</em>, 2021
<br>
<a href="https://github.com/QiyaoWei/Distributed-CMC">Code</a>
/
<p>Extend original CMC to multi-gpu training</p>
</td>
</tr>
<p>
One ICLR 2022 submission under review!
</p>
<p>
One patent coming soon!
</p>
</tbody></table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr>
<td style="padding:20px;width:100%;vertical-align:middle">
<heading>Presentations</heading>
<p>
<i>This is a collection of presentations I have made in the past</i>
</p>
<p>
- <a href="https://virtual-acml.org/conference/agenda">Here</a> is a presentation I gave at ACML 2021.
</p>
<p>
- <a href="data/misc/mirror_descent.pdf">Here</a> is technical report on mirror descent.
<a href="data/misc/mirror_descent_presentation.pdf">Here</a> is the presentation I gave to
Professor Yinyu Ye.
<!-- Representative papers are <span class="highlight">highlighted</span>.-->
</p>
<p>
- <a href="data/misc/epidemic_report.pdf">Here</a> is a technical report on epidemic control
with RL.
<a href="data/misc/epidemic_presentation.pdf">Here</a> is the presentation I gave to our
research group at Vector.
</p>
<p>
- <a href="data/misc/rce.pdf">Here</a> is a short presentation on <a
href="https://arxiv.org/pdf/2103.12656.pdf">RCE</a>
</p>
<p>
- <a href="data/misc/NPT.pdf">Here</a> is a short presentation on <a
href="https://arxiv.org/pdf/2106.02584.pdf">NPT</a>
</p>
<p>
- <a href="data/misc/IFM.pdf">Here</a> is a short presentation on <a
href="https://arxiv.org/pdf/2106.11230.pdf">IFM</a>
</p>
</td>
</tr>
</tbody>
</table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr>
<td style="padding:20px;width:100%;vertical-align:middle">
<heading>Favorite Quotes</heading>
<p>
<i>This is where I record some of the witty words I came across.
For a total stranger, they will require context to make sense. I always welcome
discussions of stuff like this.</i>
</p>
<p>
- Lex Fridman podcast with Yann LeCun "Why does the wind blow? Because the leaves move"
</p>
<p>
- E.T.Jaynes page 103 "how we solve a problem when it becomes too difficult"
</p>
<p>
- Von Neumann "You should call it entropy, for two reasons. In the first place your
uncertainty function has been used in statistical mechanics under that name, so it already
has a name. In the second place, and more important, nobody knows what entropy really is, so
in a debate you will always have the advantage"
</p>
<p>
- Paul Dirac "Experiment can only be used to check agreement with the instantiation of an idea, rather than the idea itself"
</p>
</td>
</tr>
</tbody>
</table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr>
<td style="padding:20px;width:100%;vertical-align:middle">
<heading>Miscellaneous</heading>
<p>
<i>There are also beautiful words that are not AI related.</i>
</p>
<p>
- The Great Gatsby and his "extraordinary gift of hope"
</p>
<p>
- Ricky Gervais "Just because you're offended doesn't mean you're right"
</p>
<p>
- <a href="data/misc/fail.txt">Top</a> <a href="data/misc/ideas.txt">secret</a>
</p>
</td>
</tr>
</tbody>
</table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr>
<td style="padding:20px;width:100%;vertical-align:middle">
<heading>What am I currently reading?</heading>
<p>
<i>Here is what I am up to</i>
</p>
<p>
<a href="https://www.cs.princeton.edu/courses/archive/fall19/cos597B/lecnotes/bookdraft.pdf">Theory
of Deep Learning</a>
</p>
<p>
<a href="https://rltheorybook.github.io/rltheorybook_AJKS.pdf">Reinforcement Learning:
Theory and Algorithms</a>
</p>
<p>
<a href="https://arxiv.org/pdf/2104.13478.pdf">Geometric Deep Learning</a>
</p>
<p>
<a href="https://arxiv.org/pdf/2001.04413.pdf">Backwards Feature Correction</a>
</p>
<p>
<a href="https://arxiv.org/pdf/2106.10165.pdf">The Principles of Deep Learning Theory</a>
</p>
</td>
</tr>
</tbody>
</table>
<table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;">
<tbody>
<tr>
<td style="padding:20px;width:100%;vertical-align:middle">
<heading>What if I am just getting started?</heading>
<p>
<i>Relax. It is always scary to step into a new subfield, let alone a new direction of
research.</i>
</p>
<p>
<a href="https://arxiv.org/pdf/1506.00019.pdf">Here</a> is a nice introductory paper.
</p>
<p>
<a href="http://rail.eecs.berkeley.edu/deeprlcourse/">Here</a> is my favorite deep
reinforcement learning course.
</p>
<p>
<a href="https://www.inference.vc/neural-tangent-kernels-some-intuition-for-kernel-gradient-descent/">Here</a>
is a great visualization of NTK, featuring one of my favorites blogs.
Though if you want to fully understand NTK, I personally find <a
href="https://arxiv.org/pdf/1904.11955.pdf">this</a> the most helpful, and you might
want to start from <a href="http://www.cs.toronto.edu/~radford/ftp/thesis.pdf">NNGP</a>
</p>
<p>
<a href="https://www.youtube.com/watch?v=SfSqLn0j10g">Here</a> is a fantastic introduction
on Bandits, and a gentle touch on concentration inequalities.
</p>
<p>
<a href="https://www.cs.ubc.ca/~nickhar/F18-531/NotesHoeffding.pdf">Here</a> is a great
derivation of Hoeffding's bound, making connection with union bounds.
</p>
<p>
<a href="https://ptolemy.berkeley.edu/projects/embedded/eecsx44/lectures/Spring2013/Picard.pdf">Here</a> is a great
derivation of Banach Fixed Point Theorem and Picard-Lindelof Theorem.
</p>
</td>
</tr>
</tbody>
</table>
<p>
This website is adapted from the fantastic <a href="https://jonbarron.info/">Jon Barron</a>
</p>
</td>
</tr>
</table>
</body>
</html>
<!-- <table style="width:100%;border:0px;border-spacing:0px;border-collapse:separate;margin-right:auto;margin-left:auto;"><tbody>-->
<!-- <tr onmouseout="nerfbake_stop()" onmouseover="nerfbake_start()">-->
<!-- <td style="padding:20px;width:25%;vertical-align:middle">-->
<!-- <div class="one">-->
<!-- <div class="two" id='nerfbake_image'><video width=100% height=100% muted autoplay loop>-->
<!-- <source src="images/nerfbake_15.mp4" type="video/mp4">-->
<!-- Your browser does not support the video tag.-->
<!-- </video></div>-->
<!-- <img src='images/nerfbake_160.png' width="160">-->
<!-- </div>-->
<!-- <script type="text/javascript">-->
<!-- function nerfbake_start() {-->
<!-- document.getElementById('nerfbake_image').style.opacity = "1";-->
<!-- }-->
<!-- function nerfbake_stop() {-->
<!-- document.getElementById('nerfbake_image').style.opacity = "0";-->
<!-- }-->
<!-- nerfbake_stop()-->
<!-- </script>-->
<!-- </td>-->
<!-- <td style="padding:20px;width:75%;vertical-align:middle">-->
<!-- <a href="http://nerf.live">-->
<!-- <papertitle>Baking Neural Radiance Fields for Real-Time View Synthesis</papertitle>-->
<!-- </a>-->
<!-- <br>-->
<!-- <a href="https://phogzone.com/">Peter Hedman</a>,-->
<!-- <a href="https://pratulsrinivasan.github.io/">Pratul Srinivasan</a>,-->
<!-- <a href="https://bmild.github.io/">Ben Mildenhall</a>,-->
<!-- <strong>Jonathan T. Barron</strong>,-->
<!-- <a href="https://www.pauldebevec.com/">Paul Debevec</a>-->
<!-- <br>-->
<!-- <em>arXiv</em>, 2021 -->
<!-- <br>-->
<!-- <a href="http://nerf.live">project page</a>-->
<!-- /-->
<!-- <a href="https://arxiv.org/abs/2103.14645">arXiv</a>-->
<!-- /-->
<!-- <a href="https://www.youtube.com/watch?v=5jKry8n5YO8">video</a>-->
<!-- /-->
<!-- <a href="https://nerf.live/#demos">demo</a>-->
<!-- <p></p>-->
<!-- <p>Baking a trained NeRF into a sparse voxel grid of colors and features lets you render it in real-time in your browser.</p>-->
<!-- </td>-->
<!-- <tr onmouseout="mipnerf_stop()" onmouseover="mipnerf_start()" bgcolor="#ffffd0">-->
<!-- <td style="padding:20px;width:25%;vertical-align:middle">-->
<!-- <div class="one">-->
<!-- <div class="two" id='mipnerf_image'><video width=100% height=100% muted autoplay loop>-->
<!-- <source src="images/mipnerf_ipe_yellow.mp4" type="video/mp4">-->
<!-- Your browser does not support the video tag.-->
<!-- </video></div>-->
<!-- <img src='images/mipnerf_ipe_yellow.png' width="160">-->
<!-- </div>-->
<!-- <script type="text/javascript">-->
<!-- function mipnerf_start() {-->
<!-- document.getElementById('mipnerf_image').style.opacity = "1";-->
<!-- }-->
<!-- function mipnerf_stop() {-->
<!-- document.getElementById('mipnerf_image').style.opacity = "0";-->
<!-- }-->
<!-- mipnerf_stop()-->
<!-- </script>-->
<!-- </td>-->
<!-- <td style="padding:20px;width:75%;vertical-align:middle">-->
<!-- <a href="http://jonbarron.info/mipnerf">-->
<!-- <papertitle>Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields</papertitle>-->
<!-- </a>-->
<!-- <br>-->
<!-- <strong>Jonathan T. Barron</strong>,-->
<!-- <a href="https://bmild.github.io/">Ben Mildenhall</a>,-->
<!-- <a href="http://matthewtancik.com/">Matthew Tancik</a>, <br>-->
<!-- <a href="https://phogzone.com/">Peter Hedman</a>,-->
<!-- <a href="http://www.ricardomartinbrualla.com/">Ricardo Martin-Brualla</a>,-->
<!-- <a href="https://pratulsrinivasan.github.io/">Pratul Srinivasan</a>-->
<!-- <br>-->
<!-- <em>arXiv</em>, 2021 -->
<!-- <br>-->
<!-- <a href="http://jonbarron.info/mipnerf">project page</a>-->
<!-- /-->
<!-- <a href="https://arxiv.org/abs/2103.13415">arXiv</a>-->
<!-- /-->
<!-- <a href="https://youtu.be/EpH175PY1A0">video</a>-->
<!-- <p></p>-->
<!-- <p>NeRF is aliased, but we can anti-alias it by casting cones and prefiltering the positional encoding function.</p>-->
<!-- </td>-->
<!-- </tr> -->
<!-- <tr onmouseout="ibrnet_stop()" onmouseover="ibrnet_start()">-->
<!-- <td style="padding:20px;width:25%;vertical-align:middle">-->
<!-- <div class="one">-->
<!-- <div class="two" id='ibrnet_image'><video width=100% height=100% muted autoplay loop>-->
<!-- <source src="images/ibrnet_after.mp4" type="video/mp4">-->
<!-- Your browser does not support the video tag.-->
<!-- </video></div>-->
<!-- <img src='images/ibrnet_before.jpg' width="160">-->
<!-- </div>-->
<!-- <script type="text/javascript">-->
<!-- function ibrnet_start() {-->
<!-- document.getElementById('ibrnet_image').style.opacity = "1";-->
<!-- }-->
<!-- function ibrnet_stop() {-->
<!-- document.getElementById('ibrnet_image').style.opacity = "0";-->
<!-- }-->
<!-- ibrnet_stop()-->
<!-- </script>-->
<!-- </td>-->
<!-- <td style="padding:20px;width:75%;vertical-align:middle">-->
<!-- <a href="https://ibrnet.github.io/">-->
<!-- <papertitle>IBRNet: Learning Multi-View Image-Based Rendering</papertitle>-->
<!-- </a>-->
<!-- <br>-->
<!-- <a href="https://www.cs.cornell.edu/~qqw/">Qianqian Wang</a>,-->
<!-- <a href="https://www.linkedin.com/in/zhicheng-wang-96116897/">Zhicheng Wang</a>,-->
<!-- <a href="https://www.kylegenova.com/">Kyle Genova</a>,-->
<!-- <a href="https://people.eecs.berkeley.edu/~pratul/">Pratul Srinivasan</a>,-->
<!-- <a href="https://scholar.google.com/citations?user=Rh9T3EcAAAAJ&hl=en">Howard Zhou</a>, <br>-->
<!-- <strong>Jonathan T. Barron</strong>, -->
<!-- <a href="http://www.ricardomartinbrualla.com/">Ricardo Martin-Brualla</a>,-->
<!-- <a href="https://www.cs.cornell.edu/~snavely/">Noah Snavely</a>, -->
<!-- <a href="https://www.cs.princeton.edu/~funk/">Thomas Funkhouser</a>-->
<!-- <br>-->
<!-- <em>CVPR</em>, 2021-->
<!-- <br>-->
<!-- <a href="https://ibrnet.github.io/">project page</a> /-->
<!-- <a href="https://github.com/googleinterns/IBRNet">code</a> / -->
<!-- <a href="https://arxiv.org/abs/2102.13090">arXiv</a>-->
<!-- <p></p>-->
<!-- <p>By learning how to pay attention to input images at render time, -->
<!-- we can amortize inference for view synthesis and reduce error rates by 15%.</p>-->
<!-- </td>-->
<!-- </tr>-->
<!-- <tr onmouseout="dualrefl_stop()" onmouseover="dualrefl_start()">-->
<!-- <td style="padding:20px;width:25%;vertical-align:middle">-->
<!-- <div class="one">-->
<!-- <div class="two" id='dualrefl_image'>-->
<!-- <img src='images/dualrefl_after.jpg' width="160"></div>-->
<!-- <img src='images/dualrefl_before.jpg' width="160">-->
<!-- </div>-->
<!-- <script type="text/javascript">-->
<!-- function dualrefl_start() {-->
<!-- document.getElementById('dualrefl_image').style.opacity = "1";-->
<!-- }-->
<!-- function dualrefl_stop() {-->
<!-- document.getElementById('dualrefl_image').style.opacity = "0";-->
<!-- }-->
<!-- dualrefl_stop()-->
<!-- </script>-->
<!-- </td>-->
<!-- <td style="padding:20px;width:75%;vertical-align:middle">-->
<!-- <a href="http://sniklaus.com/dualref">-->
<!-- <papertitle>Learned Dual-View Reflection Removal</papertitle>-->
<!-- </a>-->
<!-- <br>-->
<!-- <a href="http://sniklaus.com/welcome">Simon Niklaus</a>,-->
<!-- <a href="https://people.eecs.berkeley.edu/~cecilia77/">Xuaner (Cecilia) Zhang</a>,-->
<!-- <strong>Jonathan T. Barron</strong>,-->
<!-- <a href="http://nealwadhwa.com">Neal Wadhwa</a>,-->
<!-- <a href="http://rahuldotgarg.appspot.com/">Rahul Garg</a>,-->
<!-- <a href="http://web.cecs.pdx.edu/~fliu/">Feng Liu</a>,-->
<!-- <a href="https://people.csail.mit.edu/tfxue/">Tianfan Xue</a>,-->
<!-- <br>-->
<!-- <em>WACV</em>, 2021-->
<!-- <br>-->
<!-- <a href="http://sniklaus.com/dualref">project page</a> /-->
<!-- <a href="https://arxiv.org/abs/2010.00702">arXiv</a>-->
<!-- <p></p>-->
<!-- <p>-->
<!-- Reflections and the things behind them often exhibit parallax, and this lets you remove reflections from stereo pairs.-->
<!-- </p>-->
<!-- </td>-->
<!-- </tr> -->
<!-- Also, consider using <a href="https://leonidk.com/">Leonid Keselman</a>'s <a href="https://github.com/leonidk/new_website">Jekyll fork</a> of this page.-->