-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
420 lines (266 loc) · 19.3 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
<!doctype html>
<html lang="en">
<head>
<!-- META -->
<meta charset="utf-8">
<meta name="robots" content="noodp">
<meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1, user-scalable=no">
<!-- PAGE TITLE -->
<title>Ken Zeng Main Page</title>
<!-- FAVICON -->
<link rel="shortcut icon" href="assets/img/favicon.png">
<!-- FONTS -->
<link href="https://fonts.googleapis.com/css?family=Abril+Fatface%7CArapey&subset=latin-ext" rel="stylesheet">
<!-- STYLESHEETS -->
<link rel="stylesheet" type="text/css" href="assets/css/plugins.css">
<link rel="stylesheet" type="text/css" href="assets/css/main.css">
</head>
<body>
<!-- PRELOADER -->
<div class="preloader">
<div class="spinner"></div>
</div>
<!-- /PRELOADER -->
<!-- IMAGE CONTAINER -->
<div class="image-container">
<div class="background-img"></div>
</div>
<!-- /IMAGE CONTAINER -->
<!-- CONTENT AREA -->
<div class="content-area">
<!-- CONTENT AREA INNER -->
<div class="content-area-inner">
<!-- INTRO -->
<section id="intro">
<!-- CONTAINER MID -->
<div class="container-mid">
<!-- ANIMATION CONTAINER -->
<div class="animation-container animation-fade-down" data-animation-delay="0">
<h1>Hi, I'm Ken Zeng,</h1>
</div>
<!-- /ANIMATION CONTAINER -->
<!-- ANIMATION CONTAINER -->
<div class="animation-container animation-fade-left" data-animation-delay="300">
<p class="subline">
I'm currently a research machine learning engineer at <b>Ataraxis AI</b>, where I develop advanced medical machine learning systems for breast cancer.
Previously, I worked as a deep learning researcher at NYU Langone Health.
</p>
</div>
<!-- /ANIMATION CONTAINER -->
<!-- ANIMATION CONTAINER -->
<div class="animation-container animation-fade-up" data-animation-delay="600">
<a href="#about" class="smooth-scroll">Learn More<i class="fa fa-angle-down" aria-hidden="true"></i></a>
</div>
<!-- /ANIMATION CONTAINER -->
</div>
<!-- /CONTAINER MID -->
</section>
<!-- /INTRO -->
<!--
<section id="about">
<h3 class="headline scroll-animated-from-right">My Story</h3>
<p class="scroll-animated-from-right">
In 2023, I graduated from NYU with masters in data science.
</p>
<p class="scroll-animated-from-right">
While at NYU, I was fortunate enough to work at Langone Health as a deep learning researcher with professor Yiqiu Shen and professor Krzysztof J. Geras in breast cancer radiologuy.
</p>
<p class="scroll-animated-from-right">
My work and experiences at Langone gave me the opportunity to join a new NYU Langone Health spinoff company <b>Ataraxis AI</b> as a founding machine learning researcher engineer.
</p>
</section> -->
<section id="about">
<h3 class="headline scroll-animated-from-right">Career Highlights</h3>
<p class="scroll-animated-from-right">
Since joining Ataraxis AI as a <b>founding machine learning research engineer</b>, I have been working on developing advanced medical machine learning for identifying at-risk patients. In particular, my research focuses on the applications of deep learning, multimodal learning and survival analysis
for histopathology.
</p>
<p>
My team gathered data from 15 hospitals and research institutions, creating one of the largest observational datasets with detailed follow-up information in the world. Using this rich dataset, we developed one of the <b>first AI-driven tests for breast cancer prognosis</b>.
</p>
<p class="scroll-animated-from-right">
My work has been presented at top breast cancer and machine learning conferences such as San Antonio Breast Cancer Symposium and NeurIPS. Additionally, they were crucial to raising <b>4+ million dollars</b> for Ataraxis AI.
<a href="https://www.biopharmatrend.com/post/996-ataraxis-ai-steps-forward-in-precision-cancer-care-with-4-million-in-seed-funding/"><b>Click here to learn more.</b></a>
</p>
</section>
<!-- ABOUT -->
<section id="about">
<h3 class="headline scroll-animated-from-right">What skills I can bring.</h3>
<p class="scroll-animated-from-right">
<b>Programming:</b> I am skilled in Python, SQL, and various other programming languages. I have extensive experience using PyTorch for deep learning, as well as Scikit-learn for developing novel machine learning systems in the field of healthcare.
</p>
<p class="scroll-animated-from-right">
<b>Natural Language Processing:</b> I've applied large language models such as BERT models to detect hate speech, extract cancer diagnoses from medical reports, and ensure patient de-identification in my projects.
</p>
<p class="scroll-animated-from-right">
<b>Machine Learning:</b> I have expertise in utilizing statistical and deep learning models across diverse tasks, ranging from predicting breast cancer risk to forecasting company defaults.
</p>
</section>
<!-- /ABOUT -->
<!-- WORK -->
<section id="work">
<h3 class="headline scroll-animated-from-right">My Latest Work.</h3>
<br>
<!-- SHOWCASE -->
<div class="showcase">
<!-- ITEM -->
<div class="item scroll-animated-from-right">
<!-- LIGHTBOX LINK -->
<a href="#" data-featherlight="#item-1-lightbox">
<!-- INFO -->
<div class="info">
<!-- CONTAINER MID -->
<div class="container-mid">
<h5>Information Extraction from Pathology Reports</h5>
<p>We explored different combinations of data augmentations to develop a system that can extract the position, grade and subtype of key malignant findings.</p>
</div>
<!-- /CONTAINER MID -->
</div>
<!-- /INFO -->
<div class="background-image" style="background-image: url(assets/img/work/NER-paper.png)"></div>
</a>
<!-- /LIGHTBOX LINK -->
<!-- LIGHTBOX -->
<div id="item-1-lightbox" class="work-lightbox">
<img class="img-responsive" src="assets/img/work/NER-paper.png" alt="image">
<h3></h3>
<p style="color: rgb(26, 24, 24);">
My team and I developed a named entity recognition (NER) system to extract key diagnostic
elements from pathology reports. We also introduce four data augmentation methods to improve
the robustness of our model. Our BERT model trained with data augmentation achieves an entity
F1-score of 0.916 on an internal test set, surpassing the BERT baseline (0.843).
For more information, check out our Github
<a href="https://github.com/nyukat/pathology_extraction"
style="color: rgb(26, 24, 24)"
onmouseover="this.style.textDecoration='underline'"
onmouseout="this.style.textDecoration='none'"
><b>here</b></a>
</p>
</div>
<!-- /LIGHTBOX -->
</div>
<!-- /ITEM -->
</div>
<!-- /SHOWCASE -->
<br>
<!-- SHOWCASE -->
<div class="showcase">
<!-- ITEM -->
<div class="item scroll-animated-from-right">
<!-- LIGHTBOX LINK -->
<a href="#" data-featherlight="#item-2-lightbox">
<!-- INFO -->
<div class="info">
<!-- CONTAINER MID -->
<div class="container-mid">
<h5>Multimodal Risk Prediction Model for Breast Cancer Recurrence</h5>
<p>Developing a novel multimodal AI-based risk prediction model for predicting breast cancer recurrence, improving upon the accuracy of traditional genomic assays.</p>
</div>
<!-- /CONTAINER MID -->
</div>
<!-- /INFO -->
<div class="background-image" style="background-image: url(assets/img/work/multimodal-simplified.png)"></div>
</a>
<!-- /LIGHTBOX LINK -->
<!-- LIGHTBOX -->
<div id="item-2-lightbox" class="work-lightbox">
<img class="img-responsive" src="assets/img/work/multimodal-ai-recurrence.png" alt="image">
<h3></h3>
<p style="color: rgb(26, 24, 24);">
My team and I trained a multimodal AI-based risk prediction model for breast cancer recurrence.
We used a combination of clinical and pathology imaging features to predict the risk of
recurrence in patients with breast cancer. Our model achieved a C-index of 0.71 on an internal
test set, surpassing the traditional risk models. For more information, check out our paper
<a href="https://arxiv.org/pdf/2410.21256v1"
style="color: rgb(26, 24, 24)"
onmouseover="this.style.textDecoration='underline'"
onmouseout="this.style.textDecoration='none'"
><b>here</b></a>
</p>
</div>
<!-- /LIGHTBOX -->
</div>
<!-- /ITEM -->
</div>
<!-- /SHOWCASE -->
<br>
<!-- SHOWCASE -->
<div class="showcase">
<!-- ITEM -->
<div class="item scroll-animated-from-right">
<!-- LIGHTBOX LINK -->
<a href="#" data-featherlight="#item-3-lightbox">
<!-- INFO -->
<div class="info">
<!-- CONTAINER MID -->
<div class="container-mid">
<h5>Squeezing performance from pathology foundation models</h5>
<p>We demonstrate that simply tuning the hyperparameters for DINOv2 leads to similar or superior performance compared to the state-of-the-art.</p>
</div>
<!-- /CONTAINER MID -->
</div>
<!-- /INFO -->
<div class="background-image" style="background-image: url(assets/img/work/foundation-model-simplified.png)"></div>
</a>
<!-- /LIGHTBOX LINK -->
<!-- LIGHTBOX -->
<div id="item-3-lightbox" class="work-lightbox">
<img class="img-responsive" src="assets/img/work/self-supervised-learning.png" alt="image">
<h3></h3>
<p style="color: rgb(26, 24, 24);">
In this paper, we demonstrate that simply tuning the hyperparameters of popular SSL method DINOv2, using a relatively small dataset, leads to similar or superior performance.
Specifically, we conduct three successive hyperparameter searches, iteratively increasing either dataset or model size while narrowing the hyperparameter search space and carrying over promising hyperparameters.
Overall, this preliminary study demonstrates the importance of hyperparameter tuning in this domain and proposes straightforward strategies to improve foundation models with additional compute and data.
</p>
</div>
<!-- /LIGHTBOX -->
</div>
<!-- /ITEM -->
</div>
<!-- /SHOWCASE -->
</section>
<!-- /WORK -->
<section id="publications">
<h3 class="headline scroll-animated-from-right">Publications</h3>
<p class="scroll-animated-from-right">
<b>Improving Information Extraction from Pathology Reports using Named Entity Recognition</b>
</p>
<p class="scroll-animated-from-right">
<b>Ken G Zeng</b>, Tarun Dutt, Jan Witowski, GV Kranthi Kiran, Frank Yeung, Michelle Kim, ... , Freya Schnabel, Linda M Pak, Yiqiu Shen, Krzysztof J Geras
</p>
<br>
<p class="scroll-animated-from-right">
<b>Squeezing performance from pathology foundation models with chained hyperparameter searches</b>
</p>
<p class="scroll-animated-from-right">
Joseph Cappadona, <b>Ken Gary Zeng</b>, Carlos Fernandez-Granda, Jan Witowski, Yann LeCun, Krzysztof J Geras
</p>
<br>
<p class="scroll-animated-from-right">
<b>Multimodal Risk Prediction Model for Breast Cancer Recurrence</b>
</p>
<p class="scroll-animated-from-right">
Jan Witowski, <b>Ken Zeng</b>, Joseph Cappadona, Jailan Elayoubi, Elena Diana Chiru, Nancy Chan, ... Adam Brufsky, Francisco J Esteva, Lajos Pusztai, Yann LeCun, Krzysztof J Geras
</p>
<br>
</section>
<!-- FOOTER -->
<section id="footer">
<p>For more details, feel free to contact me at <b>kengaryzeng@gmail.com</b>. You can also reach out to me on Twitter or LinkedIn.</p>
<ul class="social-icons scroll-animated-from-right">
<li><a href="https://twitter.com/KenZeng56288846"><i class="fa fa-twitter" aria-hidden="true"></i></a></li>
<li><a href="https://github.com/kenzeng24"><i class="fa fa-github" aria-hidden="true"></i></a></li>
<li><a href="https://www.linkedin.com/in/ken-zeng-1b4b0a14b/"><i class="fa fa-linkedin" aria-hidden="true"></i></a></li>
<li><a href="https://www.instagram.com/ken.zeng1"><i class="fa fa-instagram" aria-hidden="true"></i></a></li>
</ul>
</section>
<!-- /FOOTER -->
</div>
<!-- /CONTENT AREA INNER -->
</div>
<!-- /CONTENT AREA -->
<!-- JAVASCRIPTS -->
<script type="text/javascript" src="assets/js/plugins.js"></script>
<script type="text/javascript" src="assets/js/main.js"></script>
</body>
</html>