International Conference on Learning Representations (ICLR) 2024

Apple is sponsoring the International Conference on Learning Representations (ICLR), which will take place in person from May 7 to 11 in Vienna, Austria. ICLR brings together professionals dedicated to the advancement of deep learning.

Accepted articles

LLM Compression: The truth is rarely pure and never simple

Ajay Jaiswal (University of Texas at Austin), Zhe Gan, Xianzhi Du, Bowen Zhang, Zhangyang Wang (University of Texas at Austin), Yinfei Yang

Data Filtering Features: Algorithmic curation of billion-scale data sets

Alex Fang (University of Washington), Albin Madappally Jose, Amit Jain, Ludwig Schmidt (University of Washington), Alexander Toshev, Vaishaal Shankar

Efficient ConvBN blocks for transfer learning and more

Kaichao You (Tsinghua University), Qin Guo (Tsinghua University), Anchang Bao (Tsinghua University), Meng Cao, Ping Huang, Jiulong Shan, Mingsheng Long (Tsinghua University)

Efficient-3Dim: Learning a novel generalizable single-image view synthesizer in a day

Yifan Jiang, Hao Tang, Rick Chang, Liangchen Song, Zhangyang Wang (University of Texas at Austin), Liangliang Cao

FedHyper: Adaptive step sizes for efficient federated learning with hypergradient descent

Ziyao Wang (University of Maryland College Park), Jianyu Wang, Ang Li (University of Maryland College Park)

FERRET: Query and ground anything anywhere and at any granularity

Haoxuan You, Haotian (AIML) Zhang, Liangliang Cao, Zhe Gan, Bowen Zhang, Zirui Wang, Xianzhi Du, Shih-Fu Chang (Columbia University), Yinfei Yang

Generative modeling with stochastic phase bridge

Tianrong Chen (Georgia tech), Jiatao Gu, Josh Susskind, Shuangfei Zhai, Laurent Dinh, Evangelos Theodorou (Georgia tech)

Instruction-based image editing guide using large multimodal language models

Tsu-Jui Fu (University of California, Santa Barbara), Wenze Hu, Xianzhi Du, William Wang (University of California, Santa Barbara), Yinfei Yang, Zhe Gan

Retrospective priorities to reward learning from human preferences

Mudit Verma (Apple Intern/Arizona State University), Rin Metcalf Susa

JointNet: Extending Text-to-Image Diffusion for Dense Distribution Modeling

Jingyang Zhang, Shiwei Li, Yuanxun Lu (Nanjing University), Tian Fang, David McKinnon, Yanghai Tsin, Long Quan (University of Science and technology), Yao Yao (Nanjing University)

Large Language Models for Generalizable Reinforcement Learning of Embedded Tasks

Andrew Szot (Georgia Institute of technology), Max Schwarzer (Université de Montréal), Harsh Agrawal, Bogdan Mazoure, Rin Metcalf Susa, Natalie Mackraz, Walter Talbott, Devon Hjelm, Alexander Toshev

Large-scale training of basic models for wearable biosignals

Salar Abbaspourazad, Oussama Elachqar, Andy Miller, Saba Emrani, Udhay Nallasamy, Ian Shapiro

LiDAR: Linear Probing Performance Detection on Co-Integration SSL Architectures

Vimal Thilak, Omid Saremi, Preetum Nakkiran, Josh Susskind, Chen Huang, Hanlin Goh, Laurent Dinh, Etai Littwin

Multiple diffusion fields

Ahmed Elhag (AIMS Senegal), Yuyang Wang, Josh Susskind, Miguel Ángel Bautista Martín

Matryoshka diffusion models

Jiatao Gu, Shuangfei Zhai, Yizhe Zhang, Josh Susskind, Navdeep Jaitly

MOFI: Learning Image Representation from Annotated Images of Noisy Entities

Wentao Wu, Aleksei Timofeev, Chen (SII) Chen, Bowen Zhang, Kun Duan, Shuangning Liu, Yantao Zheng, Jonathon Shlens (Google (contributions while at Apple)), Xianzhi Du, Zhe Gan, Yinfei Yang

Overcoming the obstacles of fitting the vision-language model for the generalization of OOD

Yuhang Zang (Nanyang Technological University), Hanlin Goh, Josh Susskind, Chen Huang

Contrastive Poly-View Learning

Amitis Shidani (Oxford University), Dan Busbridge, Devon Hjelm, Jason Ramapuram, Eeshan Gunesh Dhekane, Russ Webb

ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models

Iman Mirzadeh, Keivan Alizadeh Vahid, Sachin Mehta, Carlo C Del Mundo, Oncel Tuzel, Golnoosh Samei, Mohammad Rastegari, Mehrdad Farajtabar

TiC-CLIP: Continuous training of CLIP models

Saurabh Garg (Carnegie Mellon University), Hadi Pour Ansari, Mehrdad Farajtabar, Sachin Mehta, Raviteja Vemulapalli, Oncel Tuzel, Vaishaal Shankar, Fartash Faghri

Disappearance of gradients in reinforcement learning based on fine-tuning of linguistic models

Noam Razin (Tel Aviv University), Hattie Zhou (University of Montreal), Preetum Nakkiran, Josh Susskind, Omid Saremi, Arwen Bradley, Vimal Thilak, Etai Littwin

What algorithms can transformers learn? A study on length generalization

Hattie Zhou (University of Montreal), Omid Saremi, Etai Littwin, Arwen Bradley, Noam Razin (Tel Aviv University), Josh Susskind, Samy Bengio, Preetum Nakkiran

Conformal prediction using regression as classification

Etash Guha (Riken AIP), Shlok Natarajan (Salesforce), Thomas Mollenhoff (Riken AIP), Emtiyaz Khan (Riken AIP), Eugene Ndiaye

How to efficiently calculate Hessian vector products?

Mathieu Dagreou (Inria), Thomas Moreau (Inria), Samuel Vaiter (CNRS), Pierre Ablin

Pay only for what is uncertain: variance-adaptive Thompson sampling

Aadi Saha, Branislav Kveton (amazon)

Pseudogeneralized dynamic view synthesis from video

Xiaoming Zhao (UIUC), Fangchang Ma, Josh Susskind, Miguel Ángel Bautista Martín, Alex Colburn, Alex Schwing

When can transformers reason with abstract symbols?

Enric Boix (MIT), Josh Susskind, Omid Saremi, Emmanuel Abbe, Etai Littwin, Samy Bengio

Work accepted in the workshop

Frequency-aware masked autoencoders for multimodal pretraining on biosignals

Ran Liu (Georgia Institute of technology), Ellen Zippi, Hadi Pour Ansari, Chris Sandino, Jingping Nie (Columbia University), Hanlin Goh, Erdrin Azemi, Ali Moin

How far are we from intelligent visual deductive reasoning?

Yizhe Zhang, Richard Bai, Ruixiang Zhang, Jiatao Gu, Shuangfei Zhai, Josh Susskind, Navdeep Jaitly

Large-scale training of basic models for wearable biosignals

Salar Abbaspourazad, Oussama Elachqar, Andy Miller, Saba Emrani, Udhay Nallasamy, Ian Shapiro

Rephrase, don't repeat: Overcoming scaling laws in data-constrained language modeling

Pratyush Maini (CMU), Skyler Seto, David Grangier, Richard Bai, Yizhe Zhang, Navdeep Jaitly

Thanks

Samy Bengio is a member of the ICLR 2024 Organizing Committee.

Samy Bengio, Miguel Angel Baptist Martin, Eugene Ndiaye and Yizhe Zhang are the ICLR 2024 area presidents.

Fartash Faghri, Enrico Fini, Devon Hjelm, Bogdan Mazoure, Wenze Hu, Rin Metcalf Susa, Vimal Thilak and Luca Zappella are ICLR 2024 reviewers.