{"id":660,"date":"2023-07-24T00:21:04","date_gmt":"2023-07-24T00:21:04","guid":{"rendered":"https:\/\/todaysainews.com\/index.php\/2023\/07\/24\/google-at-icml-2023-google-research-blog\/"},"modified":"2025-04-27T07:33:00","modified_gmt":"2025-04-27T07:33:00","slug":"google-at-icml-2023-google-research-blog","status":"publish","type":"post","link":"https:\/\/todaysainews.com\/index.php\/2023\/07\/24\/google-at-icml-2023-google-research-blog\/","title":{"rendered":"Google at ICML 2023 \u2013 Google Research Blog"},"content":{"rendered":"

[ad_1]
\n<\/p>\n

\nGroups across Google actively pursue research in the field of machine learning (ML), ranging from theory and application. We build ML systems to solve deep scientific and engineering challenges in areas of language, music, visual processing, algorithm development, and more. We aim to build a more collaborative ecosystem with the broader ML research community through open-sourcing tools and datasets, publishing our work, and actively participating in conferences.\n<\/p>\n

\nGoogle is proud to be a Diamond Sponsor<\/a> of the 40th International Conference on Machine Learning<\/a> (ICML 2023), a premier annual conference, which is being held this week in Honolulu, Hawaii. As a leader in ML research, Google has a strong presence at this year\u2019s conference with over 120 accepted papers and active involvement in a number of workshops and tutorials. Google is also proud to be a Platinum Sponsor for both the LatinX in AI<\/a> and Women in Machine Learning<\/a> workshops. We look forward to sharing some of our extensive ML research and expanding our partnership with the broader ML research community.\n<\/p>\n

\nRegistered for ICML 2023? We hope you\u2019ll visit the Google booth to learn more about the exciting work, creativity, and fun that goes into solving a portion of the field\u2019s most interesting challenges. Visit the @GoogleAI<\/a> Twitter account to find out about Google booth activities (e.g., demos and Q&A sessions). See Google DeepMind\u2019s blog<\/a> to learn about their technical participation at ICML 2023.\n<\/p>\n

\nTake a look below to learn more about the Google research being presented at ICML 2023 (Google affiliations in bold<\/strong>).<\/p>\n

\n

\nScaling Vision Transformers to 22 Billion Parameters<\/a> (see blog post<\/a>)
\n
Mostafa Dehghani<\/em><\/strong>, Josip Djolonga<\/em><\/strong>, Basil Mustafa<\/em><\/strong>, Piotr Padlewski<\/em><\/strong>, Jonathan Heek<\/em><\/strong>, Justin Gilmer<\/em><\/strong>, Andreas Steiner<\/em><\/strong>, Mathilde Caron<\/em><\/strong>, Robert Geirhos<\/em><\/strong>, Ibrahim Alabdulmohsin<\/em><\/strong>, Rodolphe Jenatton<\/em><\/strong>, Lucas Beyer<\/em><\/strong>, Michael Tschannen<\/em><\/strong>, Anurag Arnab<\/em><\/strong>, Xiao Wang<\/em><\/strong>, Carlos Riquelme<\/em><\/strong>, Matthias Minderer<\/em><\/strong>, Joan Puigcerver,<\/em> Utku Evci<\/em><\/strong>, Manoj Kumar<\/em><\/strong>, Sjoerd van Steenkiste<\/em><\/strong>, Gamaleldin F. Elsayed<\/em><\/strong>, Aravindh Mahendran<\/em><\/strong>, Fisher Yu<\/em><\/strong>, Avital Oliver<\/em><\/strong>, Fantine Huot<\/em><\/strong>, Jasmijn Bastings<\/em><\/strong>, Mark Patrick Collier<\/em><\/strong>, Alexey Gritsenko<\/em><\/strong>, Vighnesh Birodkar<\/em><\/strong>, Cristina Vasconcelos<\/em><\/strong>, Yi Tay<\/em><\/strong>, Thomas Mensink<\/em><\/strong>, Alexander Kolesnikov<\/em><\/strong>, Filip Paveti\u0107<\/em><\/strong>, Dustin Tran<\/em><\/strong>, Thomas Kipf<\/em><\/strong>, Mario Lu\u010di\u0107<\/em><\/strong>, Xiaohua Zhai<\/em><\/strong>, Daniel Keysers<\/em><\/strong>, Jeremiah Harmsen<\/em><\/strong>, Neil Houlsby<\/em>
<\/strong>\n<\/p>\n

\nFast Inference from Transformers via Speculative Decoding<\/a>
\n
Yaniv Leviathan<\/em><\/strong>, Matan Kalman<\/em><\/strong>, Yossi Matias<\/em><\/strong>\n<\/p>\n

\n Best of Both Worlds Policy Optimization<\/a>
\n
Christoph Dann<\/em><\/strong>, Chen-Yu Wei<\/em>, Julian Zimmert<\/em>
<\/strong>\n<\/p>\n

Inflow, Outflow, and Reciprocity in Machine Learning<\/a>
\n
Mukund Sundararajan<\/em><\/strong>, Walid Krichene<\/em>
<\/strong>\n<\/p>\n

\n Transformers Learn In-Context by Gradient Descent<\/a>
\n
Johannes von Oswald<\/em><\/strong>, Eyvind Niklasson<\/em><\/strong>, Ettore Randazzo<\/em><\/strong>, Jo\u00e3o Sacramento<\/em>, Alexander Mordvintsev<\/em><\/strong>, Andrey Zhmoginov<\/em><\/strong>, Max Vladymyrov<\/em>
<\/strong>\n<\/p>\n

Arithmetic Sampling: Parallel Diverse Decoding for Large Language Models<\/a>
\n
Luke Vilnis<\/em><\/strong>, Yury Zemlyanskiy<\/em><\/strong>, Patrick Murray*<\/em>, Alexandre Passos*<\/em>, Sumit Sanghai<\/em>
<\/strong>\n<\/p>\n

Differentially Private Hierarchical Clustering with Provable Approximation Guarantees<\/a> (see blog post<\/a>)
\n
Jacob Imola<\/em>*, Alessandro Epasto<\/em><\/strong>, Mohammad Mahdian<\/em><\/strong>, Vincent Cohen-Addad<\/em><\/strong>, Vahab Mirrokni<\/em>
<\/strong>\n<\/p>\n

Multi-Epoch Matrix Factorization Mechanisms for Private Machine Learning<\/a>
\n
Christopher A. Choquette-Choo<\/em><\/strong>, H. Brendan McMahan<\/em><\/strong>, Keith Rush<\/em><\/strong>, Abhradeep Thakurta<\/em>
<\/strong>\n<\/p>\n

Random Classification Noise Does Not Defeat All Convex Potential Boosters Irrespective of Model Choice<\/a>
\n
Yishay Mansour<\/em><\/strong>, Richard Nock<\/em><\/strong>, Robert Williamson<\/em><\/p>\n

Simplex Random Features<\/a>
\n
Isaac Reid<\/em>, Krzysztof Choromanski<\/em><\/strong>, Valerii Likhosherstov<\/em>, Adrian Weller<\/em><\/p>\n

Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding<\/a>
\n
Kenton Lee<\/em><\/strong>, Mandar Joshi<\/em><\/strong>, Iulia Turc<\/em>, Hexiang Hu<\/em><\/strong>, Fangyu Liu<\/em>, Julian Eisenschlos<\/em><\/strong>, Urvashi Khandelwal<\/em><\/strong>, Peter Shaw<\/em><\/strong>, Ming-Wei Chang<\/em><\/strong>, Kristina Toutanova<\/em><\/strong>\n<\/p>\n

Mu2<\/sup>SLAM: Multitask, Multilingual Speech and Language Models<\/a>
\n
Yong Cheng<\/em><\/strong>, Yu Zhang<\/em><\/strong>, Melvin Johnson<\/em><\/strong>, Wolfgang Macherey<\/em><\/strong>, Ankur Bapna<\/em>
<\/strong>\n<\/p>\n

Robust Budget Pacing with a Single Sample<\/a>
\n
Santiago Balseiro<\/em><\/strong>, Rachitesh Kumar<\/em>*, Vahab Mirrokni<\/em><\/strong>, Balasubramanian Sivan<\/em><\/strong>, Di Wang<\/em>
<\/strong>\n<\/p>\n

A Statistical Perspective on Retrieval-Based Models<\/a>
\n
Soumya Basu<\/em><\/strong>, Ankit Singh Rawat<\/em><\/strong>, Manzil Zaheer<\/em><\/p>\n

Approximately Optimal Core Shapes for Tensor Decompositions<\/a>
\n
Mehrdad Ghadiri<\/em>, Matthew Fahrbach<\/em><\/strong>, Gang Fu<\/em><\/strong>, Vahab Mirrokni<\/em>
<\/strong>\n<\/p>\n

Efficient List-Decodable Regression Using Batches<\/a>
\n
Abhimanyu Das<\/em><\/strong>, Ayush Jain<\/em>*, Weihao Kong<\/em><\/strong>, Rajat Sen<\/em>
<\/strong>\n<\/p>\n

Efficient Training of Language Models Using Few-Shot Learning<\/a>
\n
Sashank J. Reddi<\/em><\/strong>, Sobhan Miryoosefi<\/em><\/strong>, Stefani Karp<\/em><\/strong>, Shankar Krishnan<\/em><\/strong>, Satyen Kale<\/em><\/strong>, Seungyeon Kim<\/em><\/strong>, Sanjiv Kumar<\/em>
<\/strong>\n<\/p>\n

Fully Dynamic Submodular Maximization Over Matroids<\/a>
\n
Paul Duetting<\/em><\/strong>, Federico Fusco<\/em>, Silvio Lattanzi<\/em><\/strong>, Ashkan Norouzi-Fard<\/em><\/strong>, Morteza Zadimoghaddam<\/em>
<\/strong>\n<\/p>\n

GFlowNet-EM for Learning Compositional Latent Variable Models<\/a>
\n
Edward J Hu<\/em>, Nikolay Malkin<\/em>, Moksh Jain<\/em>, Katie Everett<\/em><\/strong>, Alexandros Graikos<\/em>, Yoshua Bengio<\/em><\/p>\n

\nImproved Online Learning Algorithms for CTR Prediction in Ad Auctions<\/a>
\n
Zhe Feng<\/em><\/strong>, Christopher Liaw<\/em><\/strong>, Zixin Zhou<\/em>\n<\/p>\n

Large Language Models Struggle to Learn Long-Tail Knowledge<\/a>
\n
Nikhil Kandpal<\/em>, Haikang Deng<\/em>, Adam Roberts<\/em><\/strong>, Eric Wallace<\/em><\/strong>, Colin Raffel<\/em>\n<\/p>\n

Multi-channel Autobidding with Budget and ROI Constraints<\/a>
\n
Yuan Deng<\/em><\/strong>, Negin Golrezaei<\/em>, Patrick Jaillet<\/em>, Jason Cheuk Nam Liang<\/em>, Vahab Mirrokni<\/em><\/strong>\n<\/p>\n

Multi-layer Neural Networks as Trainable Ladders of Hilbert Spaces<\/a>
\n
Zhengdao Chen<\/em><\/strong>\n<\/p>\n

On User-Level Private Convex Optimization<\/a>
\n
Badih Ghazi<\/em><\/strong>, Pritish Kamath<\/em><\/strong>, Ravi Kumar<\/em><\/strong>, Raghu Meka<\/em><\/strong>, Pasin Manurangsi<\/em><\/strong>, Chiyuan Zhang<\/em><\/strong>\n<\/p>\n

PAC Generalization via Invariant Representations<\/a>
\n
Advait U Parulekar<\/em>, Karthikeyan Shanmugam<\/em><\/strong>, Sanjay Shakkottai<\/em><\/p>\n

Regularization and Variance-Weighted Regression Achieves Minimax Optimality in Linear MDPs: Theory and Practice<\/a>
\n
Toshinori Kitamura<\/em>, Tadashi Kozuno<\/em>, Yunhao Tang<\/em>, Nino Vieillard<\/em><\/strong>, Michal Valko<\/em>, Wenhao Yang<\/em>, Jincheng Mei<\/em><\/strong>, Pierre Menard<\/em>, Mohammad Gheshlaghi Azar<\/em>, Remi Munos<\/em>, Olivier Pietquin<\/em><\/strong>, Matthieu Geist<\/em><\/strong>,Csaba Szepesvari<\/em>, Wataru Kumagai<\/em>, Yutaka Matsuo<\/em>\n<\/p>\n

Speeding Up Bellman Ford via Minimum Violation Permutations<\/a>
\n
Silvio Lattanzi<\/em><\/strong>, Ola Svensson<\/em>, Sergei Vassilvitskii<\/em><\/strong>\n<\/p>\n

Statistical Indistinguishability of Learning Algorithms<\/a>
\n
Alkis Kalavasis<\/em>, Amin Karbasi<\/em><\/strong>, Shay Moran<\/em><\/strong>, Grigoris Velegkas<\/em>\n<\/p>\n

Test-Time Adaptation with Slot-Centric Models<\/a>
\n
Mihir Prabhudesai<\/em>, Anirudh Goyal<\/em>, Sujoy Paul<\/em><\/strong>, Sjoerd van Steenkiste<\/em><\/strong>, Mehdi S. M. Sajjadi<\/em><\/strong>, Gaurav Aggarwal<\/em><\/strong>, Thomas Kipf<\/em><\/strong>, Deepak Pathak<\/em>, Katerina Fragkiadaki><\/em>\n<\/p>\n

Algorithms for Bounding Contribution for Histogram Estimation Under User-Level Privacy<\/a>
\n
Yuhan Liu<\/em>*, Ananda Theertha Suresh<\/em><\/strong>, Wennan Zhu<\/em><\/strong>, Peter Kairouz<\/em><\/strong>, Marco Gruteser<\/em><\/strong>\n<\/p>\n

Bandit Online Linear Optimization with Hints and Queries<\/a>
\n
Aditya Bhaskara<\/em>, Ashok Cutkosky<\/em>, Ravi Kumar<\/em><\/strong>, Manish Purohit<\/em><\/strong>\n<\/p>\n

CLUTR: Curriculum Learning via Unsupervised Task Representation Learning<\/a>
\n
Abdus Salam Azad<\/em>, Izzeddin Gur<\/em><\/strong>, Jasper Emhoff<\/em>, Nathaniel Alexis<\/em>, Aleksandra Faust<\/em><\/strong>, Pieter Abbeel<\/em>, Ion Stoica<\/em>\n<\/p>\n

CSP: Self-Supervised Contrastive Spatial Pre-training for Geospatial-Visual Representations<\/a>
\n
Gengchen Mai<\/em>, Ni Lao<\/em><\/strong>, Yutong He<\/em>, Jiaming Song<\/em>, Stefano Ermon<\/em>\n<\/p>\n

Ewald-Based Long-Range Message Passing for Molecular Graphs<\/a>
\n
Arthur Kosmala<\/em>, Johannes Gasteiger<\/em><\/strong>, Nicholas Gao<\/em>, Stephan G\u00fcnnemann<\/em>\n<\/p>\n

Fast (1+\u03b5)-Approximation Algorithms for Binary Matrix Factorization<\/a>
\n
Ameya Velingker<\/em><\/strong>, Maximilian V\u00f6tsch<\/em>, David Woodruff<\/em><\/strong>, Samson Zhou<\/em>\n<\/p>\n

Federated Linear Contextual Bandits with User-Level Differential Privacy<\/a>
\n
Ruiquan Huang<\/em>, Huanyu Zhang<\/em>, Luca Melis<\/em>, Milan Shen<\/em>, Meisam Hejazinia<\/em><\/strong>, Jing Yang<\/em>\n<\/p>\n

Investigating the Role of Model-Based Learning in Exploration and Transfer<\/a>
\n
Jacob C Walker<\/em>, Eszter V\u00e9rtes<\/em>, Yazhe Li<\/em>, Gabriel Dulac-Arnold<\/em><\/strong>, Ankesh Anand<\/em>, Theophane Weber<\/em>, Jessica B Hamrick<\/em>\n<\/p>\n

Label Differential Privacy and Private Training Data Release<\/a>
\n
Robert Busa-Fekete<\/em><\/strong>, Andres Munoz<\/em><\/strong>, Umar Syed<\/em><\/strong>, Sergei Vassilvitskii<\/em><\/strong>\n<\/p>\n

Lifelong Language Pretraining with Distribution-Specialized Experts<\/a>
\n
Wuyang Chen<\/em>*, Yanqi Zhou<\/em><\/strong>, Nan Du<\/em><\/strong>, Yanping Huang<\/em><\/strong>, James Laudon<\/em><\/strong>, Zhifeng Chen<\/em><\/strong>, Claire Cui<\/em><\/strong>\n<\/p>\n

Multi-User Reinforcement Learning with Low Rank Rewards<\/a>
\n
Dheeraj Mysore Nagaraj<\/em><\/strong>, Suhas S Kowshik<\/em>, Naman Agarwal<\/em><\/strong>, Praneeth Netrapalli<\/em><\/strong>, Prateek Jain<\/em><\/strong>\n<\/p>\n

Multi-View Masked World Models for Visual Robotic Manipulation<\/a>
\n
Younggyo Seo<\/em>, Junsu Kim<\/em>, Stephen James<\/em>, Kimin Lee<\/em><\/strong>, Jinwoo Shin<\/em>, Pieter Abbeel<\/em>\n<\/p>\n

PaLM-E: An Embodied Multimodal Language Model<\/a> (see blog post<\/a>)
\n
Danny Driess<\/em><\/strong>, Fei Xia<\/em><\/strong>, Mehdi S. M. Sajjadi<\/em><\/strong>, Corey Lynch<\/em><\/strong>, Aakanksha Chowdhery<\/em><\/strong>, Brian Ichter<\/em><\/strong>,Ayzaan Wahid<\/em><\/strong>, Jonathan Tompson<\/em><\/strong>, Quan Vuong<\/em><\/strong>, Tianhe Yu<\/em><\/strong>, Wenlong Huang<\/em><\/strong>, Yevgen Chebotar<\/em><\/strong>, Pierre Sermanet<\/em><\/strong>, Daniel Duckworth<\/em><\/strong>, Sergey Levine<\/em><\/strong>, Vincent Vanhoucke<\/em><\/strong>, Karol Hausman<\/em><\/strong>, Marc Toussaint<\/em>, Klaus Greff<\/em><\/strong>, Andy Zeng<\/em><\/strong>, Igor Mordatch<\/em><\/strong>, Pete Florence<\/em><\/strong>\n<\/p>\n

Private Federated Learning with Autotuned Compression<\/a>
\n
Enayat Ullah<\/em>*, Christopher A. Choquette-Choo<\/em><\/strong>, Peter Kairouz<\/em><\/strong>, Sewoong Oh<\/em><\/strong>\n<\/p>\n

Refined Regret for Adversarial MDPs with Linear Function Approximation<\/a>
\n
Yan Dai<\/em>, Haipeng Luo<\/em>, Chen-Yu Wei<\/em>, Julian Zimmert<\/em><\/strong>\n<\/p>\n

Scaling Up Dataset Distillation to ImageNet-1K with Constant Memory<\/a>
\n
Justin Cui<\/em>, Ruoche Wan<\/em>, Si Si<\/em><\/strong>, Cho-Jui Hsieh<\/em>\n<\/p>\n

SGD with AdaGrad Stepsizes: Full Adaptivity with High Probability to Unknown Parameters, Unbounded Gradients and Affine Variance<\/a>
\n
Amit Attia<\/em>, Tomer Koren<\/em><\/strong>\n<\/p>\n

The Statistical Benefits of Quantile Temporal-Difference Learning for Value Estimation<\/a>
\n
Mark Rowland<\/em>, Yunhao Tang<\/em>, Clare Lyle<\/em>, R\u00e9mi Munos<\/em>, Marc G. Bellemare<\/em><\/strong>, Will Dabney<\/em>\n<\/p>\n

Unveiling The Mask of Position-Information Pattern Through the Mist of Image Features<\/a>
\n
Chieh Hubert Lin<\/em>, Hung-Yu Tseng<\/em>, Hsin-Ying Lee<\/em>, Maneesh Kumar Singh<\/em>, Ming-Hsuan Yang<\/em><\/strong>\n<\/p>\n

User-Level Private Stochastic Convex Optimization with Optimal Rates<\/a>
\n
Raef Bassily<\/em>, Ziteng Sun<\/em><\/strong>\n<\/p>\n

A Simple Zero-Shot Prompt Weighting Technique to Improve Prompt Ensembling in Text-Image Models<\/a>
\n
James Urquhart Allingham<\/em>*, Jie Ren<\/em><\/strong>, Michael W Dusenberry<\/em><\/strong>, Xiuye Gu<\/em><\/strong>, Yin Cui<\/em><\/strong>, Dustin Tran<\/em><\/strong>, Jeremiah Zhe Liu<\/em><\/strong>, Balaji Lakshminarayanan<\/em><\/strong>\n<\/p>\n

Can Large Language Models Reason About Program Invariants?<\/a>
\n
Kexin Pei<\/em><\/strong>, David Bieber<\/em><\/strong>, Kensen Shi<\/em><\/strong>, Charles Sutton<\/em><\/strong>, Pengcheng Yin<\/em><\/strong>\n<\/p>\n

\n Concurrent Shuffle Differential Privacy Under Continual Observation<\/a>
\n
Jay Tenenbaum<\/em><\/strong>, Haim Kaplan<\/em><\/strong>, Yishay Mansour<\/em><\/strong>, Uri Stemmer<\/em><\/strong>\n<\/p>\n

Constant Matters: Fine-Grained Error Bound on Differentially Private Continual Observation<\/a>
\n
Hendrik Fichtenberger<\/em><\/strong>, Monika Henzinger<\/em>, Jalaj Upadhyay<\/em>\n<\/p>\n

Cross-Entropy Loss Functions: Theoretical Analysis and Applications<\/a>
\n
Anqi Mao<\/em>, Mehryar Mohri<\/em><\/strong>, Yutao Zhong<\/em>\n<\/p>\n

Efficient Rate Optimal Regret for Adversarial Contextual MDPs Using Online Function Approximation<\/a>
\n
Orin Levy<\/em>, Alon Cohen<\/em><\/strong>, Asaf Cassel<\/em>, Yishay Mansour<\/em><\/strong>\n<\/p>\n

Fairness in Streaming Submodular Maximization Over a Matroid Constraint<\/a>
\n
Marwa El Halabi<\/em>, Federico Fusco<\/em>, Ashkan Norouzi-Fard<\/em><\/strong>, Jakab Tardos<\/em><\/strong>, Jakub Tarnawski<\/em>\n<\/p>\n

The Flan Collection: Designing Data and Methods for Effective Instruction Tuning<\/a> (see blog post<\/a>)
\n
Shayne Longpre<\/em>, Le Hou<\/em><\/strong>, Tu Vu<\/em><\/strong>, Albert Webson<\/em><\/strong>, Hyung Won Chung<\/em><\/strong>, Yi Tay<\/em><\/strong>, Denny Zhou<\/em><\/strong>, Quoc V Le<\/em><\/strong>, Barret Zoph<\/em><\/strong>, Jason Wei<\/em><\/strong>, Adam Roberts<\/em><\/strong>\n<\/p>\n

Graph Reinforcement Learning for Network Control via Bi-level Optimization<\/a>
\n
Daniele Gammelli<\/em>, James Harrison<\/em><\/strong>, Kaidi Yang<\/em>, Marco Pavone<\/em>, Filipe Rodrigues<\/em>, Francisco C. Pereira<\/em>\n<\/p>\n

Learning-Augmented Private Algorithms for Multiple Quantile Release<\/a>
\n
Mikhail Khodak<\/em>*, Kareem Amin<\/em><\/strong>, Travis Dick<\/em><\/strong>, Sergei Vassilvitskii<\/em><\/strong>\n<\/p>\n

LegendreTron: Uprising Proper Multiclass Loss Learning<\/a>
\n
Kevin H Lam<\/em>, Christian Walder<\/em><\/strong>, Spiridon Penev<\/em>, Richard Nock<\/em><\/strong>\n<\/p>\n

Measuring the Impact of Programming Language Distribution<\/a>
\n
Gabriel Orlanski<\/em>*, Kefan Xiao<\/em><\/strong>, Xavier Garcia<\/em><\/strong>, Jeffrey Hui<\/em><\/strong>, Joshua Howland<\/em><\/strong>, Jonathan Malmaud<\/em><\/strong>, Jacob Austin<\/em><\/strong>, Rishabh Singh<\/em><\/strong>, Michele Catasta<\/em>*\n<\/p>\n

Multi-task Differential Privacy Under Distribution Skew<\/a>
\n
Walid Krichene<\/em><\/strong>, Prateek Jain<\/em><\/strong>, Shuang Song<\/em><\/strong>, Mukund Sundararajan<\/em><\/strong>, Abhradeep Thakurta<\/em><\/strong>, Li Zhang<\/em><\/strong>\n<\/p>\n

Muse: Text-to-Image Generation via Masked Generative Transformers<\/a>
\n
Huiwen Chang<\/em><\/strong>, Han Zhang<\/em><\/strong>, Jarred Barber<\/em><\/strong>, AJ Maschinot<\/em><\/strong>, Jos\u00e9 Lezama<\/em><\/strong>, Lu Jiang<\/em><\/strong>, Ming-Hsuan Yang<\/em><\/strong>, Kevin Murphy<\/em><\/strong>, William T. Freeman<\/em><\/strong>, Michael Rubinstein<\/em><\/strong>, Yuanzhen Li<\/em><\/strong>, Dilip Krishnan<\/em><\/strong>\n<\/p>\n

On the Convergence of Federated Averaging with Cyclic Client Participation<\/a>
\n
Yae Jee Cho<\/em>, Pranay Sharma<\/em>, Gauri Joshi<\/em>, Zheng Xu<\/em><\/strong>, Satyen Kale<\/em><\/strong>, Tong Zhang<\/em><\/strong>\n<\/p>\n

Optimal Stochastic Non-smooth Non-convex Optimization Through Online-to-Non-convex Conversion<\/a>
\n
Ashok Cutkosky<\/em>, Harsh Mehta<\/em><\/strong>, Francesco Orabona<\/em>\n<\/p>\n

Out-of-Domain Robustness via Targeted Augmentations<\/a>
\n
Irena Gao<\/em>, Shiori Sagawa<\/em>, Pang Wei Koh<\/em><\/strong>, Tatsunori Hashimoto<\/em>, Percy Liang<\/em>\n<\/p>\n

Polynomial Time and Private Learning of Unbounded Gaussian Mixture Models<\/a>
\n
Jamil Arbas<\/em>, Hassan Ashtiani<\/em>, Christopher Liaw<\/em><\/strong>\n<\/p>\n

Pre-computed Memory or On-the-Fly Encoding? A Hybrid Approach to Retrieval Augmentation Makes the Most of Your Compute<\/a>
\n
Michiel de Jong<\/em>, Yury Zemlyanskiy<\/em><\/strong>, Nicholas FitzGerald<\/em><\/strong>, Joshua Ainslie<\/em><\/strong>, Sumit Sanghai<\/em><\/strong>, Fei Sha<\/em><\/strong>, William W. Cohen<\/em><\/strong>\n<\/p>\n

Scalable Adaptive Computation for Iterative Generation<\/a>
\n
Allan Jabri<\/em>*, David J. Fleet<\/em><\/strong>, Ting Chen<\/em><\/strong>\n<\/p>\n

Scaling Spherical CNNs<\/a>
\n
Carlos Esteves<\/em><\/strong>, Jean-Jacques Slotine<\/em>, Ameesh Makadia<\/em><\/strong>\n<\/p>\n

STEP: Learning N:M Structured Sparsity Masks from Scratch with Precondition<\/a>
\n
Yucheng Lu<\/em>, Shivani Agrawal<\/em><\/strong>, Suvinay Subramanian<\/em><\/strong>, Oleg Rybakov<\/em><\/strong>, Christopher De Sa<\/em>, Amir Yazdanbakhsh<\/em>\n<\/p>\n

Stratified Adversarial Robustness with Rejection<\/a>
\n
Jiefeng Chen<\/em>, Jayaram Raghuram<\/em>, Jihye Choi<\/em>, Xi Wu<\/em><\/strong>, Yingyu Liang<\/em>, Somesh Jha<\/em>\n<\/p>\n

When Does Privileged information Explain Away Label Noise?<\/a>
\n
Guillermo Ortiz-Jimenez<\/em>*, Mark Collier<\/em><\/strong>, Anant Nawalgaria<\/em><\/strong>, Alexander D’Amour<\/em><\/strong>, Jesse Berent<\/em><\/strong>, Rodolphe Jenatton<\/em><\/strong>, Effrosyni Kokiopoulou<\/em><\/strong>\n<\/p>\n

Adaptive Computation with Elastic Input Sequence<\/a>
\n
Fuzhao Xue<\/em>*, Valerii Likhosherstov<\/em><\/strong>, Anurag Arnab<\/em><\/strong>, Neil Houlsby<\/em><\/strong>, Mostafa Dehghani<\/em><\/strong>, Yang You<\/em>\n<\/p>\n

Can Neural Network Memorization Be Localized?<\/a>
\n
Pratyush Maini<\/em>, Michael C. Mozer<\/em><\/strong>, Hanie Sedghi<\/em><\/strong>, Zachary C. Lipton<\/em>, J. Zico Kolter<\/em>, Chiyuan Zhang<\/em><\/strong>\n<\/p>\n

Controllability-Aware Unsupervised Skill Discovery<\/a>
\n
Seohong Park<\/em>, Kimin Lee<\/em><\/strong>, Youngwoon Lee<\/em>, Pieter Abbeel<\/em>\n<\/p>\n

Efficient Learning of Mesh-Based Physical Simulation with Bi-Stride Multi-Scale Graph Neural Network<\/a>
\n
Yadi Cao<\/em>, Menglei Chai<\/em><\/strong>, Minchen Li<\/em>, Chenfanfu Jiang<\/em>\n<\/p>\n

Federated Heavy Hitter Recovery Under Linear Sketching<\/a>
\n
Adria Gascon<\/em><\/strong>, Peter Kairouz<\/strong><\/em>, Ziteng Sun<\/em><\/strong>, Ananda Theertha Suresh<\/em><\/strong>\n<\/p>\n

Graph Generative Model for Benchmarking Graph Neural Networks<\/a>
\n
Minji Yoon<\/em>, Yue Wu<\/em>, John Palowitch<\/em><\/strong>, Bryan Perozzi<\/em><\/strong>, Russ Salakhutdinov<\/em>\n<\/p>\n

H-Consistency Bounds for Pairwise Misranking Loss Surrogates<\/a>
\n
Anqi Mao<\/em>, Mehryar Mohri<\/em><\/strong>, Yutao Zhong<\/em>\n<\/p>\n

Improved Regret for Efficient Online Reinforcement Learning with Linear Function Approximation<\/a>
\n
Uri Sherman<\/em>, Tomer Koren<\/em><\/strong>, Yishay Mansour<\/em><\/strong>\n<\/p>\n

Invariant Slot Attention: Object Discovery with Slot-Centric Reference Frames<\/a>
\n
Ondrej Biza<\/em>*, Sjoerd van Steenkiste<\/em><\/strong>, Mehdi S. M. Sajjadi<\/em><\/strong>, Gamaleldin Fathy Elsayed<\/em><\/strong>, Aravindh Mahendran<\/em><\/strong>, Thomas Kipf<\/em><\/strong>\n<\/p>\n

Multi-task Off-Policy Learning from Bandit Feedback<\/a>
\n
Joey Hong<\/em>, Branislav Kveton<\/em>, Manzil Zaheer<\/em>, Sumeet Katariya<\/em>, Mohammad Ghavamzadeh<\/em><\/strong>\n<\/p>\n

Optimal No-Regret Learning for One-Sided Lipschitz Functions<\/a>
\n
Paul Duetting<\/em><\/strong>, Guru Guruganesh<\/em><\/strong>, Jon Schneider<\/em><\/strong>, Joshua Ruizhi Wang<\/em><\/strong>\n<\/p>\n

Policy Mirror Ascent for Efficient and Independent Learning in Mean Field Games<\/a>
\n
Batuhan Yardim<\/em>, Semih Cayci<\/em>, Matthieu Geist<\/em><\/strong>, Niao He<\/em>\n<\/p>\n

Regret Minimization and Convergence to Equilibria in General-Sum Markov Games<\/a>
\n
Liad Erez<\/em>, Tal Lancewicki<\/em>, Uri Sherman<\/em>, Tomer Koren<\/em><\/strong>, Yishay Mansour<\/em><\/strong>\n<\/p>\n

Reinforcement Learning Can Be More Efficient with Multiple Rewards<\/a>
\n
Christoph Dann<\/em><\/strong>, Yishay Mansour<\/em><\/strong>, Mehryar Mohri<\/em><\/strong>\n<\/p>\n

Reinforcement Learning with History-Dependent Dynamic Contexts<\/a>
\n
Guy Tennenholtz<\/em><\/strong>, Nadav Merlis<\/em>, Lior Shani<\/em><\/strong>, Martin Mladenov<\/em><\/strong>, Craig Boutlier<\/em><\/strong>\n<\/p>\n

User-Defined Event Sampling and Uncertainty Quantification in Diffusion Models for Physical Dynamical Systems<\/a>
\n
Marc Anton Finzi<\/em>*, Anudhyan Boral<\/em><\/strong>, Andrew Gordon Wilson<\/em>, Fei Sha<\/em><\/strong>, Leonardo Zepeda-Nunez<\/em><\/strong>\n<\/p>\n

Discrete Key-Value Bottleneck<\/a>
\n
Frederik Tr\u00e4uble<\/em>, Anirudh Goyal<\/em>, Nasim Rahaman<\/em>, Michael Curtis Mozer<\/em><\/strong>, Kenji Kawaguchi<\/em>, Yoshua Bengio<\/em>, Bernhard Sch\u00f6lkopf<\/em>\n<\/p>\n

DSGD-CECA: Decentralized SGD with Communication-Optimal Exact Consensus Algorithm<\/a>
\n
Lisang Ding<\/em>, Kexin Jin<\/em>, Bicheng Ying<\/em><\/strong>, Kun Yuan<\/em>, Wotao Yin<\/em>\n<\/p>\n

Exphormer: Sparse Transformers for Graphs<\/a>
\n
Hamed Shirzad<\/em>, Ameya Velingker<\/em><\/strong>, Balaji Venkatachalam<\/em><\/strong>, Danica J. Sutherland<\/em>, Ali Kemal Sinop<\/em><\/strong>\n<\/p>\n

Fast, Differentiable and Sparse Top-k: A Convex Analysis Perspective<\/a>
\n
Michael Eli Sander<\/em>*, Joan Puigcerver<\/em><\/strong>, Josip Djolonga<\/em><\/strong>, Gabriel Peyr\u00e9<\/em>, Mathieu Blondel<\/em><\/strong>\n<\/p>\n

Improved Policy Evaluation for Randomized Trials of Algorithmic Resource Allocation<\/a>
\n
Aditya Mate<\/em><\/strong>, Bryan Wilder<\/em>, Aparna Taneja<\/em><\/strong>, Milind Tambe<\/em><\/strong>\n<\/p>\n

In Search for a Generalizable Method for Source Free Domain Adaptation<\/a>
\n
Malik Boudiaf<\/em>*, Tom Denton<\/em><\/strong>, Bart van Merrienboer<\/em><\/strong>, Vincent Dumoulin<\/em><\/strong>, Eleni Triantafillou<\/em><\/strong>\n<\/p>\n

Learning Rate Schedules in the Presence of Distribution Shift<\/a>
\n
Matthew Fahrbach<\/em><\/strong>, Adel Javanmard<\/em><\/strong>, Vahab Mirrokni<\/em><\/strong>, Pratik Worah<\/em><\/strong>\n<\/p>\n

Not All Semantics Are Created Equal: Contrastive Self-Supervised Learning with Automatic Temperature Individualization<\/a>
\n
Zi-Hao Qiu<\/em>, Quanqi Hu<\/em>, Zhuoning Yuan<\/em>, Denny Zhou<\/em><\/strong>, Lijun Zhang<\/em>, Tianbao Yang<\/em>\n<\/p>\n

On the Relationship Between Explanation and Prediction: A Causal View<\/a>
\n
Amir-Hossein Karimi<\/em>*, Krikamol Muandet<\/em>, Simon Kornblith<\/em><\/strong>, Bernhard Sch\u00f6lkopf<\/em>, Been Kim<\/em><\/strong>\n<\/p>\n

On the Role of Attention in Prompt-Tuning<\/a>
\n
Samet Oymak<\/em>, Ankit Singh Rawat<\/em><\/strong>, Mahdi Soltanolkotabi<\/em>, Christos Thrampoulidis<\/em>\n<\/p>\n

PLay: Parametrically Conditioned Layout Generation Using Latent Diffusion<\/a>
\n
Chin-Yi Cheng<\/em><\/strong>, Forrest Huang<\/em><\/strong>, Gang Li<\/em><\/strong>, Yang Li<\/em><\/strong>\n<\/p>\n

The Power of Learned Locally Linear Models for Nonlinear Policy Optimization<\/a>
\n
Daniel Pfrommer<\/em>, Max Simchowitz<\/em>, Tyler Westenbroek<\/em>, Nikolai Matni<\/em>, Stephen Tu<\/em><\/strong>\n<\/p>\n

Relevant Walk Search for Explaining Graph Neural Networks<\/a>
\n
Ping Xiong<\/em>, Thomas Schnake<\/em>, Michael Gastegger<\/em>, Gr\u00e9goire Montavon<\/em>, Klaus Robert Muller<\/em><\/strong>,Shinichi Nakajima<\/em>\n<\/p>\n

Repository-Level Prompt Generation for Large Language Models of Code<\/a>
\n
Disha Shrivastava<\/em>, Hugo Larochelle<\/em><\/strong>, Daniel Tarlow<\/em><\/strong>\n<\/p>\n

Robust and Private Stochastic Linear Bandits<\/a>
\n
Vasileios Charisopoulos<\/em>*, Hossein Esfandiari<\/em><\/strong>, Vahab Mirrokni<\/em><\/strong>\n<\/p>\n

Simple Diffusion: End-to-End Diffusion for High Resolution Images<\/a>
\n
Emiel Hoogeboom<\/em><\/strong>, Jonathan Heek<\/em><\/strong>, Tim Salimans<\/em><\/strong>\n<\/p>\n

Tied-Augment: Controlling Representation Similarity Improves Data Augmentation<\/a>
\n
Emirhan Kurtulus<\/em>, Zichao Li<\/em><\/strong>, Yann Dauphin<\/em><\/strong>, Ekin D. Cubuk<\/em><\/strong>\n<\/p>\n

Why Is Public Pre-Training Necessary for Private Model Training?<\/a>
\n
Arun Ganesh<\/em><\/strong>, Mahdi Haghifam<\/em>*, Milad Nasr<\/em><\/strong>, Sewoong Oh<\/em><\/strong>, Thomas Steinke<\/em><\/strong>, Om Thakkar<\/em><\/strong>, Abhradeep Guha Thakurta<\/em><\/strong>, Lun Wang<\/em><\/strong>\n<\/p>\n

A Connection Between One-Step RL and Critic Regularization in Reinforcement Learning<\/a>
\n
Benjamin Eysenbach<\/em><\/strong>, Matthieu Geist<\/em><\/strong>, Sergey Levine<\/em><\/strong>, Ruslan Salakhutdinov<\/em>\n<\/p>\n

Beyond Uniform Lipschitz Condition in Differentially Private Optimization<\/a>
\n
Rudrajit Das<\/em>*, Satyen Kale<\/em><\/strong>, Zheng Xu<\/em><\/strong>, Tong Zhang<\/em><\/strong>, Sujay Sanghavi<\/em>\n<\/p>\n

Efficient Graph Field Integrators Meet Point Clouds<\/a>
\n
Krzysztof Choromanski<\/em><\/strong>, Arijit Sehanobish<\/em>, Han Lin<\/em>, Yunfan Zhao<\/em>, Eli Berger,<\/em> Tetiana Parshakova<\/em>, Alvin Pan<\/em>, David Watkins<\/em>, Tianyi Zhang<\/em>, Valerii Likhosherstov<\/em>, Somnath Basu Roy Chowdhury<\/em>, Avinava Dubey<\/em><\/strong>, Deepali Jain<\/em><\/strong>, Tamas Sarlos<\/em><\/strong>, Snigdha Chaturvedi<\/em>, Adrian Weller<\/em>\n<\/p>\n

Fast as CHITA: Neural Network Pruning with Combinatorial Optimization<\/a>
\n
Riade Benbaki<\/em>, Wenyu Chen<\/em>, Xiang Meng<\/em>, Hussein Hazimeh<\/em><\/strong>, Natalia Ponomareva<\/em><\/strong>, Zhe Zhao<\/em><\/strong>, Rahul Mazumder<\/em>\n<\/p>\n

Jump-Start Reinforcement Learning<\/a> (see blog post<\/a>)
\n
Ikechukwu Uchendu<\/em>*, Ted Xiao<\/em><\/strong>, Yao Lu<\/em><\/strong>, Banghua Zhu<\/em>, Mengyuan Yan<\/em>, Jos\u00e9phine Simon<\/em>, Matthew Bennice<\/em>, Chuyuan Fu<\/em>, Cong Ma<\/em>, Jiantao Jiao<\/em>, Sergey Levine<\/em><\/strong>, Karol Hausman<\/em><\/strong>\n<\/p>\n

Learning in POMDPs is Sample-Efficient with Hindsight Observability<\/a>
\n
Jonathan Lee<\/em>, Alekh Agarwal<\/em><\/strong>, Christoph Dann<\/em><\/strong>, Tong Zhang<\/em>\n<\/p>\n

Low-Variance Gradient Estimation in Unrolled Computation Graphs with ES-Single<\/a>
\n
Paul Vicol<\/em><\/strong>\n<\/p>\n

Masked Trajectory Models for Prediction, Representation, and Control<\/a>
\n
Philipp Wu<\/em>, Arjun Majumdar<\/em>, Kevin Stone<\/em>, Yixin Lin<\/em>, Igor Mordatch<\/em><\/strong>, Pieter Abbeel<\/em>, Aravind Rajeswaran<\/em>\n<\/p>\n

Overcoming Simplicity Bias in Deep Networks Using a Feature Sieve<\/a>
\n
Rishabh Tiwari<\/em><\/strong>, Pradeep Shenoy<\/em><\/strong>\n<\/p>\n

Pairwise Ranking Losses of Click-Through Rates Prediction for Welfare Maximization in Ad Auctions<\/a>
\n
Boxiang Lyu<\/em>, Zhe Feng<\/em><\/strong>, Zachary Robertson<\/em>, Sanmi Koyejo<\/em><\/strong>\n<\/p>\n

Predictive Flows for Faster Ford-Fulkerson<\/a>
\n
Sami Davies<\/em>, Benjamin Moseley<\/em>, Sergei Vassilvitskii<\/em><\/strong>, Yuyan Wang<\/em><\/strong>\n<\/p>\n

Scaling Laws for Multilingual Neural Machine Translation<\/a>
\n
Patrick Fernandes<\/em><\/strong>, Behrooz Ghorbani<\/em><\/strong>, Xavier Garcia<\/em><\/strong>, Markus Freitag<\/em><\/strong>, Orhan Firat<\/em><\/strong>\n<\/p>\n

Sequential Monte Carlo Learning for Time Series Structure Discovery<\/a>
\n
Feras Saad<\/em><\/strong>, Brian Patton<\/em><\/strong>, Matthew Douglas Hoffman<\/em><\/strong>, Rif A. Saurous<\/em><\/strong>, Vikash Mansinghka<\/em><\/strong>\n<\/p>\n

Stochastic Gradient Succeeds for Bandits<\/a>
\n
Jincheng Mei<\/em><\/strong>, Zixin Zhong<\/em>, Bo Dai<\/em><\/strong>, Alekh Agarwal<\/em><\/strong>, Csaba Szepesvari<\/em>, Dale Schuurmans<\/em><\/strong>\n<\/p>\n

Subset-Based Instance Optimality in Private Estimation<\/a>
\n
Travis Dick<\/em><\/strong>, Alex Kulesza<\/em><\/strong>, Ziteng Sun<\/em><\/strong>, Ananda Theertha Suresh<\/em><\/strong>\n<\/p>\n

The Unreasonable Effectiveness of Few-Shot Learning for Machine Translation<\/a>
\n
Xavier Garcia<\/em>, Yamini Bansal<\/em>, Colin Cherry<\/em><\/strong>, George Foster<\/em><\/strong>, Maxim Krikun<\/em><\/strong>, Melvin Johnson<\/em>, Orhan Firat<\/em><\/p>\n<\/div>\n