Skip to yearly menu bar Skip to main content


Workshop

Workshop on Machine Learning Safety

Dan Hendrycks ⋅ Victoria Krakovna ⋅ Dawn Song ⋅ Jacob Steinhardt ⋅ Nicholas Carlini
Project Page [ Contact: mlsafetyworkshop2022@gmail.com ]

Abstract

Video

Chat is not available.

Schedule

Timezone: America/Los_Angeles
7:00 AM
Opening Remarks
8:25 AM
Coffee Break
10:00 AM
Lunch
11:45 AM
Coffee Break
1:55 PM
Closing Remarks
Formalizing the Problem of Side Effect Regularization
Alex Turner ⋅ Aseem Saxena ⋅ Prasad Tadepalli
Measuring Robustness with Black-Box Adversarial Attack using Reinforcement Learning
Soumyendu Sarkar ⋅ Sajad Mousavi ⋅ Ashwin Ramesh Babu ⋅ Vineet Gundecha ⋅ Sahand Ghorbanpour ⋅ Alexander Shmakov
Investigating causal understanding in LLMs
Marius Hobbhahn ⋅ Tom Lieberum ⋅ David Seiler
Reflection Mechanisms as an Alignment Target: A Survey
Marius Hobbhahn ⋅ Eric Landgrebe ⋅ Elizabeth Barnes
Interpolating Compressed Parameter Subspaces
Siddhartha Datta ⋅ Nigel Shadbolt
Probabilistically Robust PAC Learning
Vinod Raman ⋅ Ambuj Tewari ⋅ UNIQUE SUBEDI
Misspecification in Inverse Reinforcement Learning
Joar Skalse ⋅ Alessandro Abate
Red-Teaming the Stable Diffusion Safety Filter
Javier Rando ⋅ Daniel Paleka ⋅ David Lindner ⋅ Lennart Heim ⋅ Florian Tramer
The Reward Hypothesis is False
Joar Skalse ⋅ Alessandro Abate
Training Time Adversarial Attack Aiming the Vulnerability of Continual Learning
Gyojin Han ⋅ Jaehyun Choi ⋅ HyeongGwon Hong ⋅ Junmo Kim
Measuring Reliability of Large Language Models through Semantic Consistency
Harsh Raj ⋅ Domenic Rosati ⋅ Subhabrata Majumdar
CUDA: Curriculum of Data Augmentation for Long-tailed Recognition
Sumyeong Ahn ⋅ Jongwoo Ko ⋅ Se-Young Yun
Certified defences hurt generalisation
Piersilvio De Bartolomeis ⋅ Jacob Clarysse ⋅ Fanny Yang ⋅ Amartya Sanyal
Context-Adaptive Deep Neural Networks via Bridge-Mode Connectivity
Nathan Drenkow ⋅ Alvin Tan ⋅ Clayton Ashcraft ⋅ Kiran Karra
Constraining Low-level Representations to Define Effective Confidence Scores
Joao Monteiro ⋅ Pau Rodriguez ⋅ Pierre-Andre Noel ⋅ Issam Hadj Laradji ⋅ David Vázquez
On the Robustness of Safe Reinforcement Learning under Observational Perturbations
ZUXIN LIU ⋅ Zijian Guo ⋅ Zhepeng Cen ⋅ Huan Zhang ⋅ Jie Tan ⋅ Bo Li ⋅ DING ZHAO
Improving Zero-shot Generalization and Robustness of Multi-modal Models
Yunhao Ge ⋅ Jie Ren ⋅ Ming-Hsuan Yang ⋅ Yuxiao Wang ⋅ Andrew Gallagher ⋅ Hartwig Adam ⋅ Laurent Itti ⋅ Balaji Lakshminarayanan ⋅ Jiaping Zhao
Dynamic Stochastic Ensemble with Adversarial Robust Lottery Ticket Subnetworks
Qi Peng ⋅ Wenlin Liu ⋅ Qin RuoXi ⋅ Libin Hou ⋅ Bin Yan ⋅ Linyuan Wang
Exploring Transformer Backbones for Heterogeneous Treatment Effect Estimation
yifan zhang ⋅ Hanlin Zhang ⋅ Zachary Lipton ⋅ Li Erran Li ⋅ Eric Xing
Bandits with Costly Reward Observations
Aaron Tucker ⋅ Caleb Biddulph ⋅ Claire Wang ⋅ Thorsten Joachims
Pre-training Robust Feature Extractor Against Clean-label Data Poisoning Attacks
Ting Zhou ⋅ Hanshu Yan ⋅ Lei LIU ⋅ Jingfeng Zhang ⋅ Bo Han
MoAT: Meta-Evaluation of Anti-Malware Trustworthiness
Sharon Lin ⋅ Marc Fyrbiak ⋅ Christof Paar
Cold Posteriors through PAC-Bayes
Konstantinos Pitas ⋅ Julyan Arbel
How Sure to Be Safe? Difficulty, Confidence and Negative Side Effects
John Burden ⋅ José Hernández-Orallo ⋅ Sean O hEigeartaigh
System Safety Engineering for Social and Ethical ML Risks: A Case Study
Edgar Jatho ⋅ Logan Mailloux ⋅ Shalaleh Rismani ⋅ Eugene Williams ⋅ Joshua Kroll
Quantifying Misalignment Between Agents
Aidan Kierans ⋅ Hananel Hazan ⋅ Shiri Dori-Hacohen
Lower Bounds on 0-1 Loss for Multi-class Classification with a Test-time Attacker
Sihui Dai ⋅ Wenxin Ding ⋅ Arjun Nitin Bhagoji ⋅ Daniel Cullina ⋅ Prateek Mittal ⋅ Ben Zhao
Fine-grain Inference on Out-of-Distribution Data with Hierarchical Classification
Randolph Linderman ⋅ Jingyang Zhang ⋅ Nathan Inkawhich ⋅ Hai Li ⋅ Yiran Chen
Indiscriminate Data Poisoning Attacks on Neural Networks
Yiwei Lu ⋅ Gautam Kamath ⋅ Yaoliang Yu
Mitigating Lies in Vision-Language Models
Junbo Li ⋅ Xianhang Li ⋅ Cihang Xie
Risk-aware Bayesian Reinforcement Learning for Cautious Exploration
Rohan Mitta ⋅ Hosein Hasanbeig ⋅ Daniel Kroening ⋅ Alessandro Abate
The Expertise Problem: Learning from Specialized Feedback
Oliver Daniels-Koch ⋅ Rachel Freedman
Cryptographic Auditing for Collaborative Learning
Hidde Lycklama ⋅ Nicolas Küchler ⋅ Alexander Viand ⋅ Emanuel Opel ⋅ Lukas Burkhalter ⋅ Anwar Hithnawi
An Adversarial Robustness Perspective on the Topology of Neural Networks
Morgane Goibert ⋅ Elvis Dohmatob ⋅ Thomas Ricatte
Towards Adversarial Purification using Denoising AutoEncoders
Dvij Kalaria ⋅ Aritra Hazra ⋅ Partha Chakrabarti
Adversarial Attacks on Transformers-Based Malware Detectors
Yash Jakhotiya ⋅ Heramb Patil ⋅ Jugal Rawlani
REAP: A Large-Scale Realistic Adversarial Patch Benchmark
Nabeel Hingun ⋅ Chawin Sitawarin ⋅ Jerry Li ⋅ David Wagner
Adversarial Policies Beat Professional-Level Go AIs
Tony Wang ⋅ Adam Gleave ⋅ Nora Belrose ⋅ Tom Tseng ⋅ Michael Dennis ⋅ Yawen Duan ⋅ Viktor Pogrebniak ⋅ Joseph Miller ⋅ Sergey Levine ⋅ Stuart Russell
A Deep Dive into Dataset Imbalance and Bias in Face Identification
Valeriia Cherepanova ⋅ Steven Reich ⋅ Samuel Dooley ⋅ Hossein Souri ⋅ John Dickerson ⋅ Micah Goldblum ⋅ Tom Goldstein
Part-Based Models Improve Adversarial Robustness
Chawin Sitawarin ⋅ Kornrapat Pongmala ⋅ Yizheng Chen ⋅ Nicholas Carlini ⋅ David Wagner
Smoothed-SGDmax: A Stability-Inspired Algorithm to Improve Adversarial Generalization
Jiancong Xiao ⋅ Jiawei Zhang ⋅ Zhiquan Luo ⋅ Asuman Ozdaglar
Hidden Poison: Machine unlearning enables camouflaged poisoning attacks
Jimmy Di ⋅ Jack Douglas ⋅ Jayadev Acharya ⋅ Gautam Kamath ⋅ Ayush Sekhari
DrML: Diagnosing and Rectifying Vision Models using Language
Yuhui Zhang ⋅ Jeff Z. HaoChen ⋅ Shih-Cheng Huang ⋅ Kuan-Chieh Wang ⋅ James Zou ⋅ Serena Yeung
Deceiving the CKA Similarity Measure in Deep Learning
MohammadReza Davari ⋅ Stefan Horoi ⋅ Amine Natik ⋅ Guillaume Lajoie ⋅ Guy Wolf ⋅ Eugene Belilovsky
A Mechanistic Lens on Mode Connectivity
Ekdeep S Lubana ⋅ Eric Bigelow ⋅ Robert Dick ⋅ David Krueger ⋅ Hidenori Tanaka
Visual Prompting for Adversarial Robustness
Aochuan Chen ⋅ Peter Lorenz ⋅ Yuguang Yao ⋅ Pin-Yu Chen ⋅ Sijia Liu
Mitigating Dataset Bias by Using Per-sample Gradient
Sumyeong Ahn ⋅ SeongYoon Kim ⋅ Se-Young Yun
A General Framework for Safe Decision Making: A Convex Duality Approach
Martino Bernasconi ⋅ Federico Cacciamani ⋅ Nicola Gatti ⋅ Francesco Trovò
A Unifying Framework for Online Safe Optimization
Matteo Castiglioni ⋅ Andrea Celli ⋅ Alberto Marchesi ⋅ Giulia Romano ⋅ Nicola Gatti
Targeted Adversarial Self-Supervised Learning
Minseon Kim ⋅ Hyeonjeong Ha ⋅ Sooel Son ⋅ Sung Ju Hwang
Canary in a Coalmine: Better Membership Inference with Ensembled Adversarial Queries
Yuxin Wen ⋅ Arpit Bansal ⋅ Hamid Kazemi ⋅ Eitan Borgnia ⋅ Micah Goldblum ⋅ Jonas Geiping ⋅ Tom Goldstein
Can Large Language Models Truly Follow your Instructions?
Joel Jang ⋅ Seonghyeon Ye ⋅ Minjoon Seo
Broken Neural Scaling Laws
Ethan Caballero ⋅ kshitij Gupta ⋅ Irina Rish ⋅ David Krueger
Do Domain Generalization Methods Generalize Well?
Akshay Mehra ⋅ Bhavya Kailkhura ⋅ Pin-Yu Chen ⋅ Jihun Hamm
What You See is What You Get: Principled Deep Learning via Distributional Generalization
Bogdan Kulynych ⋅ Yao-Yuan Yang ⋅ Yaodong Yu ⋅ Jaroslaw Blasiok ⋅ Preetum Nakkiran
Adversarial poisoning attacks on reinforcement learning-driven energy pricing
Sam Gunn ⋅ Doseok Jang ⋅ Orr Paradise ⋅ Lucas Spangher ⋅ Costas J Spanos
OOD Detection with Class Ratio Estimation
Mingtian Zhang ⋅ Andi Zhang ⋅ Tim Xiao ⋅ Yitong Sun ⋅ Steven McDonagh
Alignment as a Dynamic Process
Paul de Font-Reaulx
The Use of Non-epistemic Values to Account for Bias in Automated Decision Making
Jesse Hoey ⋅ Gabrielle Chan ⋅ Mathieu Doucet ⋅ Christopher Risi ⋅ Freya Zhang
Few-Shot Transferable Robust Representation Learning via Bilevel Attacks
Minseon Kim ⋅ Hyeonjeong Ha ⋅ Sung Ju Hwang
What 'Out-of-distribution' Is and Is Not
Sebastian Farquhar ⋅ Yarin Gal
On Outlier Exposure with Generative Models
Konstantin Kirchheim ⋅ Frank Ortmeier
Steering Large Language Models using APE
Yongchao Zhou ⋅ Andrei Muresanu ⋅ Ziwen Han ⋅ Keiran Paster ⋅ Silviu Pitis ⋅ Harris Chan ⋅ Jimmy Ba
Error Resilient Deep Neural Networks using Neuron Gradient Statistics
Chandramouli Amarnath ⋅ Abhijit Chatterjee ⋅ Kwondo Ma ⋅ Mohamed Mejri
Aligning Robot Representations with Humans
Andreea Bobu ⋅ Andi Peng ⋅ Pulkit Agrawal ⋅ Julie A Shah ⋅ Anca Dragan
Instance-Aware Observer Network for Out-of-Distribution Object Segmentation
Victor Besnier ⋅ Andrei Bursuc ⋅ Alexandre Briot ⋅ David Picard
A general framework for reward function distances
Erik Jenner ⋅ Joar Skalse ⋅ Adam Gleave
Certifiable Robustness Against Patch Attacks Using an ERM Oracle
Kevin Stangl ⋅ Avrim Blum ⋅ Omar Montasser ⋅ Saba Ahmadi
On the Adversarial Robustness of Vision Transformers
Rulin Shao ⋅ Zhouxing Shi ⋅ Jinfeng Yi ⋅ Pin-Yu Chen ⋅ Cho-Jui Hsieh
Unified Probabilistic Neural Architecture and Weight Ensembling Improves Model Robustness
Sumegha Premchandar ⋅ Sanket Jantre ⋅ Prasanna Balaprakash ⋅ Sandeep Madireddy
All’s Well That Ends Well: Avoiding Side Effects with Distance-Impact Penalties
Charlie Griffin ⋅ Joar Skalse ⋅ Lewis Hammond ⋅ Alessandro Abate
System III: Learning with Domain Knowledge for Safety Constraints
Fazl Barez ⋅ Hosein Hasanbeig ⋅ Alessandro Abate
Can Active Sampling Reduce Causal Confusion in Offline Reinforcement Learning?
Gunshi Gupta ⋅ Tim G. J. Rudner ⋅ Rowan McAllister ⋅ Adrien Gaidon ⋅ Yarin Gal
Two-Turn Debate Does Not Help Humans Answer Hard Reading Comprehension Questions
Alicia Parrish ⋅ Harsh Trivedi ⋅ Nikita Nangia ⋅ Jason Phang ⋅ Vishakh Padmakumar ⋅ Amanpreet Singh Saimbhi ⋅ Samuel Bowman
Panning for Gold in Federated Learning: Targeted Text Extraction under Arbitrarily Large-Scale Aggregation
Hong-Min Chu ⋅ Jonas Geiping ⋅ Liam Fowl ⋅ Micah Goldblum ⋅ Tom Goldstein
Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
Kevin Wang ⋅ Alexandre Variengien ⋅ Arthur Conmy ⋅ Buck Shlegeris ⋅ Jacob Steinhardt
From plane crashes to algorithmic harm: applicability of safety engineering frameworks for responsible ML
Shalaleh Rismani ⋅ Renee Shelby ⋅ Andrew Smart ⋅ Edgar Jatho ⋅ Joshua Kroll ⋅ AJung Moon ⋅ Negar Rostamzadeh
Best of Both Worlds: Towards Adversarial Robustness with Transduction and Rejection
Nils Palumbo ⋅ Yang Guo ⋅ Xi Wu ⋅ Jiefeng Chen ⋅ Yingyu Liang ⋅ Somesh Jha
c-MBA: Adversarial Attack for Cooperative MARL Using Learned Dynamics Model
Nhan H Pham ⋅ Lam Nguyen ⋅ Jie Chen ⋅ Thanh Lam Hoang ⋅ Subhro Das ⋅ Lily Weng
Adversarial Attacks on Feature Visualization Methods
Michael Eickenberg ⋅ Eugene Belilovsky ⋅ Jonathan Marty
On The Fragility of Learned Reward Functions
Lev McKinney ⋅ Yawen Duan ⋅ David Krueger ⋅ Adam Gleave
Decepticons: Corrupted Transformers Breach Privacy in Federated Learning for Language Models
Liam Fowl ⋅ Jonas Geiping ⋅ Steven Reich ⋅ Yuxin Wen ⋅ Wojciech Czaja ⋅ Micah Goldblum ⋅ Tom Goldstein
Adversarial Robustness for Tabular Data through Cost and Utility Awareness
Klim Kireev ⋅ Bogdan Kulynych ⋅ Carmela Troncoso
Epistemic Side Effects & Avoiding Them (Sometimes)
Toryn Klassen ⋅ Parand Alizadeh Alamdari ⋅ Sheila McIlraith
Improving the Robustness of Conditional Language Models by Detecting and Removing Input Noise
Kundan Krishna ⋅ Yao Zhao ⋅ Jie Ren ⋅ Balaji Lakshminarayanan ⋅ Jiaming Luo ⋅ Mohammad Saleh ⋅ Peter Liu
Diagnostics for Deep Neural Networks with Automated Copy/Paste Attacks
Stephen Casper ⋅ Kaivalya Hariharan ⋅ Dylan Hadfield-Menell
On Representation Learning Under Class Imbalance
Ravid Shwartz-Ziv ⋅ Micah Goldblum ⋅ Yucen Li ⋅ C. Bayan Bruss ⋅ Andrew Gordon Wilson
DP-InstaHide: Data Augmentations Provably Enhance Guarantees Against Dataset Manipulations
Eitan Borgnia ⋅ Jonas Geiping ⋅ Valeriia Cherepanova ⋅ Liam Fowl ⋅ Arjun Gupta ⋅ Amin Ghiasi ⋅ Furong Huang ⋅ Micah Goldblum ⋅ Tom Goldstein
Geometric attacks on batch normalization
Amur Ghose ⋅ Apurv Gupta ⋅ Yaoliang Yu ⋅ Pascal Poupart
Improving Adversarial Robustness via Joint Classification and Multiple Explicit Detection Classes
Sina Baharlouei ⋅ Fatemeh Sheikholeslami ⋅ Meisam Razaviyayn ⋅ J. Zico Kolter
On the Abilities of Mathematical Extrapolation with Implicit Models
Juliette Decugis ⋅ Alicia Tsai ⋅ Ashwin Ganesh ⋅ Max Emerling ⋅ Laurent El Ghaoui
Netflix and Forget: Fast Severance From Memorizing Training Data in Recommendations
Xinlei XU ⋅ Jiankai Sun ⋅ Xin Yang ⋅ Yuanshun Yao ⋅ Chong Wang
Evaluating Worst Case Adversarial Weather Perturbations Robustness
Yihan Wang ⋅ Yunhao Ba ⋅ Howard Zhang ⋅ Huan Zhang ⋅ Achuta Kadambi ⋅ Stefano Soatto ⋅ Alex Wong ⋅ Cho-Jui Hsieh
BAAT: Towards Sample-specific Backdoor Attack with Clean Labels
Yiming Li ⋅ Mingyan Zhu ⋅ Chengxiao Luo ⋅ Haiqing Weng ⋅ Yong Jiang ⋅ Tao Wei ⋅ Shu-Tao Xia
Out-Of-Distribution Detection Is Not All You Need
Joris Guerin ⋅ Kevin Delmas ⋅ Raul S Ferreira ⋅ Jérémie Guiochet
Revisiting Robustness in Graph Machine Learning
Lukas Gosch ⋅ Daniel Sturm ⋅ Simon Geisler ⋅ Stephan Günnemann
Assistance with large language models
Dmitrii Krasheninnikov ⋅ Egor Krasheninnikov ⋅ David Krueger
Image recognition time for humans predicts adversarial vulnerability for models
David Mayo ⋅ Jesse Cummings ⋅ Xinyu Lin ⋅ Boris Katz ⋅ Andrei Barbu
Unifying Grokking and Double Descent
Xander Davies ⋅ Lauro Langosco ⋅ David Krueger
Assessing Robustness of Image Recognition Models to Changes in the Computational Environment
Nikolaos Louloudakis ⋅ Perry Gibson ⋅ José Cano ⋅ Ajitha Rajan
Fake It Until You Make It : Towards Accurate Near-Distribution Novelty Detection
Hossein Mirzaei ⋅ Mohammadreza Salehi ⋅ Sajjad Shahabi ⋅ Efstratios Gavves ⋅ Cees Snoek ⋅ Mohammad Sabokrou ⋅ Mohammad Hossein Rohban