Mengye Ren @mengyer Twitter profile | Pikagi

Pikagi

Mengye Ren

@mengyer

3,790

Followers

722

Following

25

Media

632

Statuses

Assistant Professor of Comp Sci. and Data Sci. at NYU. Machine Learning, Computer Vision, Human-like AI.

New York, NY

https://t.co/ElSH9iOTwH

Joined July 2009

Don't wanna be here? Send us removal request.

Pinned Tweet

@mengyer

Mengye Ren

5 months

Humans and animals learn visual knowledge through continuous streams of experiences. How do we perform unsupervised continual learning (UCL) in the wild? Yipeng's latest paper reveals three essential components for UCL success in real-world scenarios: Plasticity, Stability, and

@yipengzz

Yipeng Zhang

5 months

What should you care about when continually improving your model’s representations with self-supervised learning? Check out our paper titled *Integrating Present and Past in Unsupervised Continual Learning* to appear at #CLVision2024 and #CoLLAs2024 ! 1/🧵

Tweet media one

2

6

20

1

4

24

Last Seen Profiles

@ZarakiBris

@JonathanSturel

@Inital_com

@CapcomIR

@sammysaiya

@_userv

@DontBeNumb

@GervasLungu

@Mikury_

@pbver2

@amnestyindo

@BavarianSoccer

@wales_walking

@gracecharisxox

@zryo47

@brassband47

@omaigayt

@mrm8488

@bokeplokalmalam

@paul_brickley

@LisaASchulz2

@luckywithlola

@18lewis67

@dashtiabataa

@bokeplokalmalam

@victorsavkin

@Jogigios

@ashley_har45753

@_Inari_Kohaku_

@EndMassMigrate

@Jason_Stollman

@ozzu_dq

@hs_visio

@EyadQunaibi1

@yahyaahmed91151

@WaterwayProject

@mengyer

Mengye Ren

1 year

Wondering about how to train deep neural networks without backprop? Check out our ICLR 2023 paper: Forward gradient computes gradient information from forward pass. But it is slow and noisy — it computes the directional gradient along a random weight

Tweet card media

Scaling Forward Gradient With Local Losses

Forward gradient learning computes a noisy directional gradient and is a biologically plausible alternative to backprop for learning deep neural networks. However, the standard forward gradient...

32

214

955

@mengyer

Mengye Ren

3 years

Thrilled to share that I'll be joining as an assistant professor at NYU Courant CS + Center for Data Science, starting in Fall 2022! @nyuniversity @CILVRatNYU @NYUDataScience . Deeply grateful to my mentors and collaborators, esp. to @zemelgroup & @RaquelUrtasun !

87

17

571

@mengyer

Mengye Ren

3 years

I am looking for incoming students at NYU Courant CS or Center for Data Sci starting fall 2022. If you have interest in working with me on topics like meta/continual/embodied/representation learning+vision, please reach out & mention me in your PhD application. Deadline is Dec 12

14

80

352

@mengyer

Mengye Ren

1 year

Dalle 3 generates watches and clocks pointing at 10:10 and fails to generate other times because almost all of the product images advertise timepieces using 10:10 since it is more visually appealing. #dalle3 #GenAI

Tweet media one

Tweet media two

@sichengzhuml

Sicheng Zhu

1 year

I doubt it 🤔

Tweet media one

1

10

57

6

36

244

@mengyer

Mengye Ren

2 years

China, a country that produces one of the most papers in AI, will be underrepresented in today’s #NeurIPS in New Orleans, due to their zero-covid policy, the most draconian in the world. While I am boarding on a plane, many Chinese are still locked in their buildings for months.

0

7

123

@mengyer

Mengye Ren

2 years

@wangzhian8848 可以查找资料、细化结构、文案润色、结合whisper做更好的字幕，翻译多种语言，gpt-4编程能力可以制作视频小工具，结合用midjourney生成图片和视频封面以及切换图画风格。

3

7

87

@mengyer

Mengye Ren

2 years

ML in the past: Loss function, convergence, learning rate, generalization, complexity, gradient, variance, normalization, duality... ML today: Calling a fine-tuning API and then a completion API. Learning how APIs work to keep myself up-to-date.

3

6

77

@mengyer

Mengye Ren

3 years

Although ConvNets are already translation-equivariant, we found that a general motion-equivariance objective actually guides the network to learn semantically meaningful representations. For more details, check out our latest #ICCV2021 paper:

@yuwen_x

Yuwen Xiong

3 years

Curious about self-supervised learning from raw videos in the wild? Our #ICCV2021 paper FlowE proposed a Flow-Equivariance objective that leverages moving objects, useful for segmentation/detection. Check out paper+video for detail: @RaquelUrtasun @mengyer

2

16

77

0

13

72

@mengyer

Mengye Ren

2 years

See y’all tmr

Tweet media one

1

0

69

@mengyer

Mengye Ren

5 years

Our NeurIPS 2019 paper "Incremental Few-Shot Learning with Attention Attractor Network" studies how to learn few-shot new concepts on top of previously learned old classes by meta-learning a regularization function. Paper: GitHub:

Tweet card media

GitHub - renmengye/inc-few-shot-attractor-public: Code for Paper "Incremental Few-Shot Learning...

Code for Paper "Incremental Few-Shot Learning with Attention Attractor Networks" - renmengye/inc-few-shot-attractor-public

1

17

68

@mengyer

Mengye Ren

9 months

🚨 New Research Alert! People have found safety training of LLMs can be easily undone through finetuning. How can we ensure safety in customized LLM finetuning while making finetuning still useful? Check out our latest work led by Jiachen Zhao! @jcz12856876 🔍 Our study reveals:

Tweet media one

Tweet media two

1

12

66

@mengyer

Mengye Ren

6 years

With 1000+ accepted paper in NIPS, it's virtually impossible to go through the whole list without questioning the meaning of life... We should have some cluster based visualization tool to help us navigate... Or, an auto generated news article highlights the latest breakthroughs.

3

4

61

@mengyer

Mengye Ren

2 years

There is no better way to burn money than doing research.

5

3

60

@mengyer

Mengye Ren

2 years

My talk at NeurIPS 2022 MetaLearn workshop on "meta-learning within a lifetime" is now available in recording:

Tweet card media

Meta Learning Within a Lifetime

Mengye Ren's talk at NeurIPS 2022 MetaLearn workshop. New Orleans, LA, USA. Dec 02, 2022.AbstractReal world agents like us learn from an online experience an...

www.youtube.com

1

8

55

@mengyer

Mengye Ren

2 years

Unsupervised learning has become more and more supervised.

3

3

55

@mengyer

Mengye Ren

10 months

Introducing LifelongMemory, an LLM-based personalized AI for egocentric video natural language query (NLQ). This amazing work is led by Ying Wang @yingwww_

1

7

57

@mengyer

Mengye Ren

2 years

When AI is used by totalitarian regimes to censor, silence, and purge people everyday, as AI researchers, what can we do and what have we done? There is a fine line to be cut, and the future of humanity depends on it.

3

1

51

@mengyer

Mengye Ren

2 years

@YiMaTweets For me, it’s time to break my rule.

0

2

47

@mengyer

Mengye Ren

2 years

Why you should care about ongoing 🇨🇳 protests?For years, 🇨🇳 has been using AI to censor msgs &posts, and prosecute ppl w/ diff political views,or ppl who just wanna end zero-covid. You should care b/c the cool things you created—transformers, CNNs—are likely held in evil’s hands.

@mengyer

Mengye Ren

2 years

China, a country that produces one of the most papers in AI, will be underrepresented in today’s #NeurIPS in New Orleans, due to their zero-covid policy, the most draconian in the world. While I am boarding on a plane, many Chinese are still locked in their buildings for months.

0

7

123

2

0

46

@mengyer

Mengye Ren

2 years

China’s zero-covid policy is causing a humanitarian catastrophe. Ppl locked in homes for 100+ days. 10 died in a fire and firefighters couldn’t get thru blockades. 27 died in a bus to covid camps. Ppl starving, ppl dying. Now protest has begun. Let’s end zero-covid, end the CCP.

0

1

44

@mengyer

Mengye Ren

2 years

Congrats to all award winners! Test-of-time award: AlexNet -- I can't believe it has been 10 years!

@NeurIPSConf

NeurIPS Conference

2 years

Happy to announce the award winners of #NeurIPS2022

3

88

399

0

4

44

@mengyer

Mengye Ren

3 years

Can we ask self-driving cars to pay attention? Our recent ICRA'21 paper proposed a learned attention mechanism that is not only more efficient but also much safer! Joint work by Bob @zengwenyuan1995 @binyangderek Ming @RaquelUrtasun

2

4

43

@mengyer

Mengye Ren

10 months

There are two clusters of bad course project reports: 1) those obviously used ChatGPT and 2) those don't even bother using ChatGPT.

0

1

43

@mengyer

Mengye Ren

2 years

May I suggest @iclr_conf to add line numbers in its template?

0

0

37

@mengyer

Mengye Ren

4 years

Humans learn new knowledge and concepts continually, incrementally, with only a few labeled examples, in a dynamically changing environment. Yet standard machine learning emphasizes "train-then-evaluate" which is inflexible to adapt once a model is trained. 1/10

2

6

34

@mengyer

Mengye Ren

4 years

Today 10:30AM ET: I will give a tea talk @MILAMontreal on "Towards continual & compositional few-shot learning". You can join livestream or watch recording here:

Welcome to Mila Tea Talks! What are tea talks ? Tea talks are scientific keynote-like talks given at Mila on Friday mornings (10:30am EDT). We either have an internal or invited speaker give a ~1...

sites.google.com

3

1

34

@mengyer

Mengye Ren

2 years

French President Macron in New Orleans

Tweet media one

0

5

32

@mengyer

Mengye Ren

2 years

Check out our ICLR 2023 workshop on Scene Representations for Autonomous Driving. We have an amazing line of speakers below. See you in Kigali, Rwanda!

@kashyap7x

Kashyap Chitta @ ECCV 🇮🇹

2 years

Excited to announce the ICLR 2022 workshop on Scene Representations for Autonomous Driving! Submission: (by 15.02.2023) We're accepting contributions in the form of blogs, repos, or PDFs. We also have a track for commenting on others' work, check it out!

Tweet media one

1

23

108

0

2

30

@mengyer

Mengye Ren

6 years

Code for our ICML 2018 paper "Learning to Reweight Examples for Robust Deep Learning" is shared here: Many thanks to co-authors and co-workers at Uber.

Tweet card media

GitHub - uber-research/learning-to-reweight-examples: Code for paper "Learning to Reweight Examples...

Code for paper "Learning to Reweight Examples for Robust Deep Learning" - uber-research/learning-to-reweight-examples

0

2

27

@mengyer

Mengye Ren

3 years

Congratulations to @nicvadivelu on receiving the Jessie W.H. Zou Memorial Award! I am honored to be his mentor at Uber and super excited to see his future trajectory as a rising star in the field!

@thegautamkamath

Gautam Kamath

@thegautamkamath

3 years

Huge congratulations to Nicholas Vadivelu ( @nicvadivelu ), an exemplar @UWCheritonCS undergrad, who just won the Jessie W.H. Zou Memorial Award for Excellence in Undergraduate Research!! Time for a #ResearchIntro 1/4

Tweet media one

1

8

90

1

2

25

@mengyer

Mengye Ren

4 years

Can we push back #neurips deadline? In the midst of this global pandemic, the least we want is to increase stress on our already fragile mental state in an endless isolation. Also #icml2020 cycle has been delayed. Show some coordination among big conference organizers?

1

1

24

@mengyer

Mengye Ren

4 years

After such rebuttal hustles, I realized that maybe it's to the best interest for both #neurips and myself to reject my papers, so I can save them for a non-virtual conference in the future.

1

0

24

@mengyer

Mengye Ren

4 years

Publishing a paper in the old days: Rounds of feedbacks/edits and eventually get accepted. Publishing a paper today: Open up the lottery ticket to see what are the scores, and withdraw and re-submit to the next conf deadline in a week.

0

0

23

@mengyer

Mengye Ren

1 year

Check out our ICML workshop on Localized Learning! We have an amazing line of speakers: @geoffreyhinton , @irinarish , Edouard Oyallon, Claudia Clopath, @timos_m , Qu Yang, and Stephen Gou! Our workshop paper deadline is May 24 (1 week after NeurIPS). See you in Hawaii!

@davidinouye

David I. Inouye

1 year

Excited to announce our ICML Workshop on Localized Learning! CfP Deadline: Wed, May 24 (1 week after NeurIPS) Awards: Free workshop registration! Organizers: @davidinouye @mengyer @MateuszOnAI @meickenberg @ebelilov Gao Huang @icmlconf #ICML2023

Tweet media one

3

8

25

1

3

22

@mengyer

Mengye Ren

2 years

For those of you who are not aware of what is happening in Shanghai, my hometown, and the rest of China.

@selinawangtv

Selina Wang

2 years

This is an extraordinary, historic moment in China Protests are breaking out across the country-from Beijing, to elite colleges, to other major cities, and even far flung places Shocking to hear people chanting for Xi to step down This is people past their breaking point @cnn

634

5K

16K

1

1

21

@mengyer

Mengye Ren

4 years

Check out our latest paper on learning novel concepts by imitating human sketches! Code is also available.

@zemelgroup

Zemel Group

4 years

Inspired by how humans use sketches to communicate novel visual concepts, our SketchEmbedNet learns to draw and can be used for few-shot learning. We also show interesting properties like linear embedding composition. arXiv: GitHub:

Tweet media one

Tweet media two

Tweet media three

Tweet media four

1

0

24

0

0

18

@mengyer

Mengye Ren

2 years

My thoughts on the latest China protests. A race between AI and humanity: Time is running out for the Chinese people and all of us.

1

1

18

@mengyer

Mengye Ren

4 years

Binged workshops for two entire days... Talk overflow in my brain🤣

1

0

18

@mengyer

Mengye Ren

2 years

Geoff Hinton weighs in on the past and potential of AI - CBS News

Tweet card media

"Godfather of artificial intelligence" weighs in on the past and potential of AI

Geoffrey Hinton, who works with Google and mentors AI's rising stars, started researching artificial intelligence over 40 years ago.

www.cbsnews.com

0

3

16

@mengyer

Mengye Ren

4 years

Giving a few-shot learning talk at MIT vision seminar today at 12PM ET!

Tweet card media

MIT Visual Computing Seminar

Check out MIT Visual Computing Seminar mailing list († Zoom link and sign-up spreadsheet) and YouTube Channel for further information.

sites.google.com

0

1

17

@mengyer

Mengye Ren

1 year

Aloha! Our ICML workshop on localized learning is happening this Saturday at room 310. See our official ICML workshop page for a detailed schedule! We have an amazing line of speakers and panelists -->

@mengyer

Mengye Ren

1 year

Check out our ICML workshop on Localized Learning! We have an amazing line of speakers: @geoffreyhinton , @irinarish , Edouard Oyallon, Claudia Clopath, @timos_m , Qu Yang, and Stephen Gou! Our workshop paper deadline is May 24 (1 week after NeurIPS). See you in Hawaii!

1

3

22

0

3

16

@mengyer

Mengye Ren

4 years

I didn't realize that COVID has been haunting us for so long until the toilet paper I hoarded since the beginning of March started to run out...

2

0

16

@mengyer

Mengye Ren

2 years

China has officially given up its notorious zero-covid policy—a huge political progress for the protesters. As China is opening up, let’s remember the people in Shanghai, Xinjiang, and many other cities have been locked in their homes for months, all for nothing.

0

1

17

@mengyer

Mengye Ren

4 years

Check out our new #ECCV2020 paper, presenting this week! End-to-end + Interpretable #SelfDrivingCars

@sergioksas

Sergio Casas

4 years

Can we learn safe #SelfDriving end-to-end while keeping interpretability? Yes! Check out our #ECCV2020 paper on neural #motionplanning through interpretable semantic representations. Joint work w/ A.Sadat, @mengyer , P.Dhawan, X.Wu, @RaquelUrtasun @UberATG

1

5

31

0

1

16

@mengyer

Mengye Ren

4 years

Official & authentic zgroup account. Stay tuned with our latest research!

@zemelgroup

Zemel Group

4 years

Hello world! 🎉 We are the Z-Group a.k.a. Rich Zemel’s lab at @UofT and @VectorInst ! Excited to join the party on Twitter. We’ll be posting about ML papers, as well as other fun things that come to mind 🧠 Check the next tweet to follow some students in our lab:

4

32

190

0

1

16

@mengyer

Mengye Ren

1 year

Hallucination or creativity?

1

0

15

@mengyer

Mengye Ren

2 years

What an eventful week. My thoughts on recent advances in AI.

1

2

14

@mengyer

Mengye Ren

4 years

Before desk rejecting papers with format issues, please provide a format checker next time to the authors in the portal. This is a new process introduced this year and it’s unfair to authors since we don’t understand which rules are flagging and there is no warning. @NeurIPSConf

2

1

14

@mengyer

Mengye Ren

2 years

#ChatGPT and other LLMs can help ESL community write better scientific papers and promote DEI in research.

0

0

13

@mengyer

Mengye Ren

5 years

This over-hyped field is morbidly competitive.

@hardmaru

hardmaru

5 years

“To start a PhD in ML, without insider referral, you need to do work equivalent to half of a PhD. Hence, in Apr 2019, I decided to dedicate all my time until Jan 2020 to publish in either NeurIPS or ICLR. If I fail, I would become a JavaScript programmer.” — @andreas_madsen ‼️

14

138

768

0

1

13

@mengyer

Mengye Ren

1 year

@jon_barron could be nice for encoding things non-color/intensity.

1

0

12

@mengyer

Mengye Ren

1 year

Why does the US still have a discriminatory immigration policy based on country of birth? Like, it's 2023, not 1882...

1

0

13

@mengyer

Mengye Ren

3 years

Why is AWS or GCP so reluctant to approve my request of launching an 8 GPU instance?

3

0

12

@mengyer

Mengye Ren

4 years

Towards more interactive #selfdriving , we propose a new motion forecasting network based on the transformer architecture to explicitly model interaction among actors. Check out our recent IROS'20 paper, available on arXiv: #SelfDrivingCars @uber @uberatg

Tweet media one

Tweet media two

Tweet media three

1

2

12

@mengyer

Mengye Ren

5 years

Wondering about self-driving vehicles encountering unseen animals on the road? Our CoRL paper studies how to recognize unknown objects in the open world. Joint work with Kelvin Wong, Shenlong Wang, Ming Liang and Raquel Urtasun.

Tweet card media

Identifying Unknown Instances for Autonomous Driving

In the past few years, we have seen great progress in perception algorithms, particular through the use of deep learning. However, most existing approaches focus on a few categories of interest,...

0

1

10

@mengyer

Mengye Ren

2 years

Most of our next generation will probably not understand why communicating with computers in human language was so non-trivial back in the days. Just like we are so used to take consumer electronics for granted.

3

0

11

@mengyer

Mengye Ren

5 years

So, cmt server is down again. It’s 2019, come on! A few thousand submission really isn’t that many!

1

0

11

@mengyer

Mengye Ren

4 years

In standard few-shot learning (FSL), an elephant is always an elephant no matter which episode it is. Check out our new paper that extends FSL to more flexible classification criteria -->

Tweet card media

Probing Few-Shot Generalization with Attributes

Despite impressive progress in deep learning, generalizing far beyond the training distribution is an important open challenge. In this work, we consider few-shot classification, and aim to shed...

@zemelgroup

Zemel Group

4 years

Check out our new paper Flexible Few-Shot Learning -- the same object can belong to different classes depending on context. We found unsupervised representation is better than supervised. A short version at NeurIPS metalearn workshop today at 10 EST.

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3

8

45

1

0

11

@mengyer

Mengye Ren

4 years

I want a customized TensorBoard.

0

0

11

@mengyer

Mengye Ren

4 years

Code & data are released:

Tweet card media

GitHub - renmengye/oc-fewshot-public: Code associated with paper "Wandering Within a World: Online...

Code associated with paper "Wandering Within a World: Online Contextualized Few-Shot Learning" - renmengye/oc-fewshot-public

@zemelgroup

Zemel Group

4 years

Our recent paper features a natural few-shot learning paradigm that continually introduce new objects in a sequence, such as when wandering through a new environment like walking room-to-room in a new house. Code & data: @mengyer

Tweet media one

Tweet media two

Tweet media three

0

2

29

0

1

10

@mengyer

Mengye Ren

2 years

Nuclear fusion will likely provide clean and nearly endless energy for massive AGIs.

1

0

10

@mengyer

Mengye Ren

2 years

Continuous vs discrete, local vs centralized, exploration vs exploitation… They are not only pairs of opposite concepts and tradeoffs, but also pairs of one concept embedded in another repeated over many layers.

1

0

10

@mengyer

Mengye Ren

5 years

ICLR score ladder designers and reviewers, score ONE, it's the lowest positive integer out of 10, do you really mean it? To me, it's like saying this paper is absolutely trash.

0

0

9

@mengyer

Mengye Ren

2 years

It is scary if an entire population uses one proprietary chat software to have daily private conversations. Switching cost is too high for individuals. For cell phone, you can switch service provider and still talk to anyone. Shouldn’t we have open protocols for our digital life?

1

1

10

@mengyer

Mengye Ren

4 years

Imagine a cloak can make you invisible? We build a car-top that can hide cars from LiDAR detectors. Let's start making self-driving safer and more secure. Work by James Tu and other amazing collaborators!

Tweet media one

@DeepAI

DeepAI

4 years

Physically Realizable Adversarial Examples for LiDAR Object Detection by James Tu et al. including @mengyer #DeepLearning #ComputerScience

0

2

4

1

0

10

@mengyer

Mengye Ren

2 years

Who will get the first PhD degree advised by GPT?

1

1

10

@mengyer

Mengye Ren

5 years

I hate anywhere on earth deadlines...

0

0

9

@mengyer

Mengye Ren

1 year

@savvyRL Maybe with a smaller LR and batch size?

1

0

5

@mengyer

Mengye Ren

3 years

Excited to see our graph hypernet (GHN) just got upgraded to be much stronger now. Check out GHN2 from Boris Knyazev et al.:

@BorisAKnyazev

Boris Knyazev

3 years

Do we still need SGD/Adam to train neural networks? Based on our #NeurIPS2021 paper, we are one step closer to replacing hand-designed optimizers with a single meta-model. Our meta-model can predict parameters for almost any neural network in just one forward pass. (1/n)

30

323

2K

0

0

8

@mengyer

Mengye Ren

2 years

In Shanghai subway, the police manually search cell phones for apps like Telegram.

@whyyoutouzhele

李老师不是你老师

@whyyoutouzhele

2 years

今晚上海地铁民警正在挨个查手机

2K

4K

14K

2

0

9

@mengyer

Mengye Ren

1 year

In 2023, the role of photography is getting redefined. Whereas AI generates near perfect images, Apple’s new Vision Pro lets us immerse in historical moments. To many people, photos are visual souvenirs that activates our memory. But how do we preserve our digital memory? A

0

2

9

@mengyer

Mengye Ren

3 years

gmail has been aggressively classifying emails as spam these days. A lot of surprise when I opened up my spam folder. Even for replies to my original email? Such a false positives rate seems like something is wrong...

1

0

8

@mengyer

Mengye Ren

11 months

Testing #GPT4V with the Stroop color-word test -- Impressive performance but not quite there. Making more sequential attentional mistakes rather than color-word interference.

Tweet media one

0

1

8

@mengyer

Mengye Ren

5 years

Tim Hortons, Canada’s #1 coffee brand, after these many years, still on their way learning how to design a coffee cup that doesn’t spill on my hand.

1

0

8

@mengyer

Mengye Ren

2 years

Rogers is getting ridiculous. How can this possibly happen for so long?

1

0

8

@mengyer

Mengye Ren

10 months

Using national security as a disguise for making discriminatory laws is unacceptable. See Science article: New Florida law blocks Chinese students from academic labs

Tweet card media

New Florida law blocks Chinese students from academic labs

Top state education officials need to approve any proposed hires after extensive vetting

www.science.org

1

1

8

@mengyer

Mengye Ren

3 years

Why is today's authentication technology a disaster? To sit down at my desk, I 1. Transfer my coffee and food to my left hand, and complete the rest of the tasks with another hand 2. Fail to unlock my iphone with a mask for 3 seconds and enter password with my right hand

2

0

7

@mengyer

Mengye Ren

2 years

Why does “sunset grill” only serve breakfast and close at 3pm? Could never get a sunset grill during sunset. Should change its name to “sunrise grill”

2

0

8

@mengyer

Mengye Ren

4 years

Check out our latest paper LoCo for local unsupervised contrastive learning . Joint work with @yuwen_xiong_cs and @RaquelUrtasun

Tweet card media

LoCo: Local Contrastive Representation Learning

Deep neural nets typically perform end-to-end backpropagation to learn the weights, a procedure that creates synchronization constraints in the weight update step across layers and is not...

@yuwen_x

Yuwen Xiong

4 years

Can we make unsupervised learning not backprop end-to-end? In our paper , we propose LoCo that matches end-to-end BP using SIMCLR, unlocking the potential for model parallelism + memory savings + bio plausibility. @mengyer @RaquelUrtasun @UberATG @Uber

Tweet media one

Tweet media two

Tweet media three

Tweet media four

0

1

21

0

0

7

@mengyer

Mengye Ren

10 months

@RichardMCNgo future does not seem great for introverts

1

0

9

@mengyer

Mengye Ren

2 years

@_jasonwei A Chinese first name is probably a better indicator if they are native native Chinese speakers.

0

0

7

@mengyer

Mengye Ren

8 months

@RogerGrosse I didn’t pass these filters, and I thank Canada for giving me an alternative path.

0

0

7

@mengyer

Mengye Ren

4 years

Just realized how boring our digital lives are: Everyone has the same GUI, same phone, same email, same calendar, same msg, same website, etc. the list goes on. The most you get is changing the bg img or colors. When will design be decoupled from tech products?

2

0

7

@mengyer

Mengye Ren

1 year

Protip: When plotting a training curve, do not include all of your 1,000,000 training step data points in a vectorized figure in the paper pdf -- it will clog up your pdf reader and your printer.

0

0

7

@mengyer

Mengye Ren

1 year

@savvyRL Yeah, but I still think there is also the question of dataset size. I don't think the usage of the word has changed much from before -- pretraining on ImageNet classification and finetuned on some other domain specific image classification dataset also meant the same learning

1

0

2

@mengyer

Mengye Ren

1 year

It seems that GPT-4 can directly edit LaTeX now.

0

0

6

@mengyer

Mengye Ren

6 years

New preprint on learning to reweight training examples w/ Wenyuan, Bin and Raquel. More robustness towards label noises and class imbalance by comparing the grad. direction of clean and noisy inputs.

Tweet card media

Learning to Reweight Examples for Robust Deep Learning

Deep neural networks have been shown to be very powerful modeling tools for many supervised learning tasks involving complex input patterns. However, they can also easily overfit to training set...

0

2

6

@mengyer

Mengye Ren

4 years

Progress I made over #quarantine : Developed a taste for cheap instant coffee. Over the past 5 months (2 cups per day), the money I spend on coffee is less than a dozen #Starbucks .

0

0

6

@mengyer

Mengye Ren

11 months

@savvyRL What's going on with all the anonymous hate comments under this thread?

5

0

6

@mengyer

Mengye Ren

6 years

TFW the reviewers say that there are minor flaws but not pointing out where are the flaws... that the idea is not novel and not SOTA but not giving the actual references... How about just have an option: "not in a good mood of accepting this paper"?

0

0

6

@mengyer

Mengye Ren

5 years

UofT: please cancel in-person classes. The Canadian government: please shut down entrance from countries undergoing outbreak. Why does Canada always have to be the slow mover, again?

0

0

6

@mengyer

Mengye Ren

5 years

Just found out COLT means a young male horse 🐴

0

0

6

@mengyer

Mengye Ren

4 years

Absolutely love the virtual paper experience. Thank you for the hard work!

@srush_nlp

Sasha Rush

4 years

1/ Spent the last couple weeks in quarantine obsessively coding a website for Virtual ICLR with @hen_str . We wanted to build something that was fun to browse, async first, and feels alive.

39

399

2K

0

0

6

@mengyer

Mengye Ren

2 years

@ylecun @MetaAI @paperswithcode Very cool work! Could Galactica potentially plagiarize existing papers, like in OpenAI’s codex that copied other people’s licensed code line-by-line? Or what if two people used the same tool and wrote two similar papers?

1

0

4

@mengyer

Mengye Ren

7 years

This blog introduces Uber ATG Toronto’s recent research on SBNet, leveraging block activation sparsity for speeding-up CNN inference. Our customized CUDA kernels are also open-sourced on GitHub (). via @ubereng

Tweet card media

GitHub - uber-research/sbnet: Sparse Blocks Networks

Sparse Blocks Networks. Contribute to uber-research/sbnet development by creating an account on GitHub.

0

0

6

@mengyer

Mengye Ren

4 years

Congrats to CVPR making into top-5 scientific publication venues according to Google Scholar

0

0

6

@mengyer

Mengye Ren

5 years

Why there is no Uber for family doctors? Why is it so hard to find a family doctor? Are we waiting for AI to fill these jobs?

1

0

5

@mengyer

Mengye Ren

4 years

Post pandemic, everyone will have a little mirror on the meeting desk to check their self-image.

0

0

5

@mengyer

Mengye Ren

4 years

Joint work with Lingyun (Luke) Li, Bin Yang, Ming Liang @zengwenyuan1995 @seanseg @RaquelUrtasun

0

2

5

@mengyer

Mengye Ren

3 years

@Eleni30fillou @hugo_larochelle @dumoulinv Congratulations Eleni! Such a great news!

1

0

5