@pinyuchenTW Profile picture

Pin-Yu Chen

@pinyuchenTW

Principal research scientist@IBM Research & Chief Scientist@RPI-IBM AI Research Collaboration & PI@MIT-IBM AI Lab. IJCAI Computers & Thought Award Winner.

Similar User
Tengyu Ma photo

@tengyuma

Trustworthy ML Initiative (TrustML) photo

@trustworthy_ml

Zico Kolter photo

@zicokolter

Aleksander Madry photo

@aleks_madry

Sharon Y. Li photo

@SharonYixuanLi

Greg Yang photo

@TheGregYang

Prof. Anima Anandkumar photo

@AnimaAnandkumar

MIT-IBM Watson AI Lab photo

@MITIBMLab

Sham Kakade photo

@ShamKakade6

Zeyuan Allen-Zhu photo

@ZeyuanAllenZhu

Quanquan Gu photo

@QuanquanGu

Chaowei Xiao photo

@ChaoweiX

Nan Jiang photo

@nanjiang_cs

Yisong Yue photo

@yisongyue

Jerry Li photo

@jerryzli

Pinned

Here are a @IBMResearch blog post summarizing our work on adversarial robustness @NeurIPSConf #NeurIPS2021 and a video describing our research and vision in this space. Check out this fun and important research topic!🎁 Blog: ibm.co/3s4BHpr Video: youtu.be/9B2jKXGUZtc

Congratulations to @PinYuChen for his outstanding work in the advancement of trusted AI. His research on adversarial robustness contributed to an unparalleled 8 papers accepted into @NeurIPSConf this year! Learn more about this work: ibm.co/3s4BHpr

Tweet Image 1


Pin-Yu Chen Reposted

🧑‍💻 The code of our NeurIPS'24 LLM safety landscape paper is now publicly available at: github.com/poloclub/llm-l… x.com/RealAnthonyPen…

LLM safety alignment can be easily compromised by finetuning with only a few adversarially designed training examples. 😲 Why? Are all open-source LLMs equally vulnerable to finetuning? How fast does the model start to break during finetuning? 🤔

Tweet Image 1


This reminds me of adversarial robustness research - only a few representative attacks like PGD/CW/query-based are long-lasting. Breaking is easy, fixing is hard. Hope we'll see more work on mitigating jailbreak attacks. E.g., check out our Gradient Cuff arxiv.org/abs/2403.00867

Jailbreaks have become a new sort of ImageNet competition instead of helping us better understand LLM security. I wrote a blogpost about what I think valuable research could look like 🧵



Pin-Yu Chen Reposted

Thrilled to receive the #AdvML Rising Star Award! 🌟 Grateful for the recognition of my research on responsible GenAI. Looking forward to presenting at @AdvMLFrontiers during @NeurIPSConf 2024! 🚀✨

Please join me in congratulating this year's #AdvML Rising Star Award winners, @AlexRobey23 & @xuandongzhao, for their research accomplishments in AI robustness and safety. Their award talks will be presented at @AdvMLFrontiers @NeurIPSConf 2024 Details: sites.google.com/view/advml/adv…

Tweet Image 1


Pin-Yu Chen Reposted

Big thanks to the award committee—@pinyuchenTW, @uiuc_aisecure, @sijialiu17, @cho_jui_hsieh—and the @AdvMLFrontiers workshop organizers! 🙏 Congrats as well to @AlexRobey23 for being the other AdvML Rising Star winner!


Pin-Yu Chen Reposted

Our team open-sourced two new models: Granite Guardian 3.0 2B and Granite Guardian 3.0 8B. Read more: linkedin.com/pulse/ibm-open… Hugging Face: huggingface.co/collections/ib… Documentation: ibm.com/granite/docs/m… Try them out!


Pin-Yu Chen Reposted

I'm grateful to have received the Adversarial ML Rising Star Award! 🚀 @AdvMLFrontiers is a fantastic venue. Many thanks to the award committee @pinyuchenTW @uiuc_aisecure @sijialiu17 @cho_jui_hsieh and to the workshop organizers!

Please join me in congratulating this year's #AdvML Rising Star Award winners, @AlexRobey23 & @xuandongzhao, for their research accomplishments in AI robustness and safety. Their award talks will be presented at @AdvMLFrontiers @NeurIPSConf 2024 Details: sites.google.com/view/advml/adv…

Tweet Image 1


Pin-Yu Chen Reposted

I can't wait to talk about our new work on jailbreaking LLM-controlled robots at @NeurIPSConf in Vancouver! x.com/AlexRobey23/st…

Chatbots like ChatGPT can be jailbroken to output harmful text. But what about robots? Can AI-controlled robots be jailbroken to perform harmful actions in the real world? Our new paper finds that jailbreaking AI-controlled robots isn't just possible. It's alarmingly easy. 🧵



Please join me in congratulating this year's #AdvML Rising Star Award winners, @AlexRobey23 & @xuandongzhao, for their research accomplishments in AI robustness and safety. Their award talks will be presented at @AdvMLFrontiers @NeurIPSConf 2024 Details: sites.google.com/view/advml/adv…

Tweet Image 1

Pin-Yu Chen Reposted

Come to our #COLM2024 poster #13 this afternoon and hear from Irene Ko on test-time estimation robustness-accuracy trade-offs in LLMs with synthetic data. Joint work with @pinyuchenTW Yung-Sung Chung Luca Daniel and myself.

Tweet Image 1

Great summary on model merging and mode connectivity. Also adding our work on 1. Mode connectivity and backdoors: openreview.net/forum?id=SJgwz… 2. Mode connectivity and adversarial examples: arxiv.org/abs/2009.02439 3. Safety loss landscape exploration for LLMs: arxiv.org/abs/2405.17374

Model merging is a popular research topic with applications to LLM alignment and specialization. But, did you know this technique has been studied since the 90s? Here’s a brief timeline… (Stage 0) Original work on model merging dates back to the 90s [1], where authors showed…

Tweet Image 1


Pin-Yu Chen Reposted

In our @kdd_news paper, @Changchang_Yin @pinyuchenTW @BingshengY @dakuowang, Jeff and I explore human-centered AI for sepsis early prediction. Join our oral talk tomorrow at 4:30 pm Room 124-125 if you attend KDD this week. @OSUengineering @OhioStateMed @OSUbigdata @OSUWexMed

An #AI tool proposed by @OhioState scientists to support decision-making about patients at risk for sepsis accounts for its lack of certainty & suggests what clinical data it needs to improve its predictive performance. bit.ly/3ABwGuC



The Adversarial Machine Learning Rising Star Awards deadline is in two weeks! Submit your application and help us promote your work and research vision! @trustworthy_ml @LLMSecurity @ml_safety @safe_paper

🚩(1/2) Please help forward the Call for the 2024 Adversarial Machine Learning (AdvML) Rising Star Awards! We promote junior researchers in AI safety, robustness, and security. Award events are hosted at AdvML'Frontiers workshop @NeurIPSConf 2024 Info: sites.google.com/view/advml/adv…

Tweet Image 1
Tweet Image 2


Pin-Yu Chen Reposted

Our follow-up work on the LLM theory---- the learning and generalization mechanism of Chain-of-Thought (CoT), will be presented in the next two days of the @icmlconf workshops. 1. Fri 26 Jul., Straus 2, HiLD Workshop. 2. Sat 27 Jul., Straus 2, TF2M Workshop.

Tweet Image 1

Are you a big fan of in-context learning (ICL)? Check out our @IBMResearch blog post highlighting our @icmlconf paper demystifying ICL. We characterize how ICL learns and generalizes. With @LiHongkang_jntm Meng Wang @rpi Songtao Lu & Xiaodong Cui Blog: research.ibm.com/blog/demystify…


Submit your work and join our workshop to explore the frontier of adversarial machine learning for AI!

📢 We're back with a new edition, this year at @NeurIPSConf in Vancouver! Paper deadline is August 30th, we are looking forward to your submissions!

Tweet Image 1


Pin-Yu Chen Reposted

The 3rd AdvML-Frontiers Workshop (@AdvMLFrontiers advml-frontier.github.io) is set for #NeurIPS 2024 (@NeurIPSConf)! This year, we're delving into the expansion of the trustworthy AI landscape, especially in large multi-modal systems. @trustworthy_ml @llm_sec🚀 We're now…

Tweet Image 1

Loading...

Something went wrong.


Something went wrong.