Proxy Model

Proxy models are surrogate models that approximate the behavior of complex systems, reducing computational costs and improving efficiency in various applications. Current research focuses on developing physics-aware and data-driven proxy models for diverse domains, including process systems, natural language processing, and reinforcement learning, often employing techniques like neural networks, Markov Decision Processes, and information-theoretic approaches. These advancements are significant because they enable faster model selection, more efficient training of large language models, improved explainability of complex systems, and enhanced privacy in federated learning settings. The resulting improvements in speed, efficiency, and explainability have broad implications across numerous scientific fields and practical applications.

Papers

October 30, 2024

Keep on Swimming: Real Attackers Only Need Partial Knowledge of a Multi-Model System
Julian Collado, Kevin Stangl
Adversarial Attack Adversarial Perturbation Multiple Model Incomplete Information Pseudo Victim AttackER Proxy Model

October 15, 2024

Adaptive Data Optimization: Dynamic Sample Selection with Scaling Laws
Yiding Jiang, Allan Zhou, Zhili Feng, Sadhika Malladi, J. Zico Kolter
Scaling Law Computational Efficiency Adaptive Optimization Data Mixture Proxy Model Computational Overhead

July 7, 2024

PAPM: A Physics-aware Proxy Model for Process Systems
Pengwei Liu, Zhongkai Hao, Xingyu Ren, Hangjie Yuan, Jiayang Ren, Dong Ni
Physic Aware Physical Prior Proxy Model Process System Physical Knowledge

June 13, 2024

ProxyLM: Predicting Language Model Performance on Multilingual Tasks via Proxy Models
David Anugraha, Genta Indra Winata, Chenyue Li, Patrick Amadeus Irawan, En-Shiun Annie Lee
Language Model Natural Language Processing Performance Prediction Transparent Proxy Server Architecture Multilingual Task Proxy Model Novel Iterative Data Enhancement

March 7, 2024

Proxy-RLHF: Decoupling Generation and Alignment in Large Language Model with Proxy
Yu Zhu, Chuxiong Sun, Wenfei Yang, Wenqiang Wei, Bo Tang, Tianzhu Zhang, Zhiyu Li, Shifeng Zhang, Feiyu Xiong, Jie Hu, Mingchuan yang
Large Language Model Reinforcement Learning Faithful Generation Alignment Problem Reinforcement Learning From Human Feedback Alignment Approach Transparent Proxy Server Architecture Proxy Model Alignment Task

February 19, 2024

Small Models, Big Insights: Leveraging Slim Proxy Models To Decide When and What to Retrieve for LLMs
Jiejun Tan, Zhicheng Dou, Yutao Zhu, Peidong Guo, Kun Fang, Ji-Rong Wen
Large Language Model DCU Insight AQ LLM Inference Question Answering Task Small Model Knowledge Acquisition Proxy Model

February 3, 2024

Generating In-Distribution Proxy Graphs for Explaining Graph Neural Networks
Zhuomin Chen, Jiaxing Zhang, Jingchao Ni, Xiaoting Li, Yuchen Bian, Md Mezbahul Islam, Ananda Mohan Mondal, Hua Wei, Dongsheng Luo
Graph Neural Network High Explainability Subgraph Explanation Graph Generator Proxy Model

January 16, 2024

November 1, 2023

Efficient Human-AI Coordination via Preparatory Language-based Convention
Cong Guan, Lichao Zhang, Chunpeng Fan, Yichen Li, Feng Chen, Lihe Li, Yunjia Tian, Lei Yuan, Yang Yu
High Efficiency Artificial General Intelligence Intelligent Agent Academic Conference Proxy Model Pre Print Human AI Coordination

October 19, 2023

Learn from the Past: A Proxy Guided Adversarial Defense Framework with Self Distillation Regularization
Yaohua Liu, Jiaxin Gao, Xianghao Jiao, Zhu Liu, Xin Fan, Risheng Liu
Adversarial Training Model Robustness Catastrophic Overfitting Proxy Model Defense Framework

October 2, 2023

LoFT: Local Proxy Fine-tuning For Improving Transferability Of Adversarial Attacks Against Large Language Model
Muhammad Ahmed Shah, Roshan Sharma, Hira Dhamyal, Raphael Olivier, Ankit Shah, Joseph Konan, Dareen Alharthi, Hazim T Bukhari, Massa Baali, Soham Deshmukh, Michael Kuhlmann, Bhiksha Raj, Rita Singh
Large Language Model Adversarial Attack Fine Tuning Task Transferability Transferable Attack Private Model Adversarial Suffix Proxy Model Proxy Tuning

March 3, 2023

Learning Stabilization Control from Observations by Learning Lyapunov-like Proxy Models
Milan Ganai, Chiaki Hirayama, Ya-Chien Chang, Sicun Gao
Reinforcement Learning Lyapunov Function Evolving Observation Expert Trajectory Proxy Model State Sequence

October 6, 2022

Weak Proxies are Sufficient and Preferable for Fairness with Missing Sensitive Attributes
Zhaowei Zhu, Yuanshun Yao, Jiankai Sun, Hang Li, Yang Liu
Procedural Fairness Sensitive Attribute Proxy Dataset Fairness Analysis Proxy Model

February 28, 2022

Enhance transferability of adversarial examples with model architecture
Mingyuan Fan, Wenzhong Guo, Shengxing Yu, Zuobin Ying, Ximeng Liu
Adversarial Attack Adversarial Example Model Architecture Improving Transferability Effective Model Proxy Model

February 6, 2022

Learning Synthetic Environments and Reward Networks for Reinforcement Learning
Fabio Ferreira, Thomas Nierhoff, Andreas Saelinger, Frank Hutter
Reinforcement Learning Reinforcement Learning Algorithm Synthetic Environment Proxy Model Interpretable Reward

November 22, 2021

Decentralized Federated Learning through Proxy Model Sharing
Shivam Kalra, Junfeng Wen, Jesse C. Cresswell, Maksims Volkovs, Hamid R. Tizhoosh
Decentralized Data Decentralized Federated Learning Transparent Proxy Server Architecture Proxy Model Model Sharing

Proxy Model

Papers

Keep on Swimming: Real Attackers Only Need Partial Knowledge of a Multi-Model System

Adaptive Data Optimization: Dynamic Sample Selection with Scaling Laws

PAPM: A Physics-aware Proxy Model for Process Systems

ProxyLM: Predicting Language Model Performance on Multilingual Tasks via Proxy Models

Proxy-RLHF: Decoupling Generation and Alignment in Large Language Model with Proxy

Small Models, Big Insights: Leveraging Slim Proxy Models To Decide When and What to Retrieve for LLMs

Generating In-Distribution Proxy Graphs for Explaining Graph Neural Networks

An Explainable Proxy Model for Multiabel Audio Segmentation

Probabilistically Robust Watermarking of Neural Networks

Efficient Human-AI Coordination via Preparatory Language-based Convention

Learn from the Past: A Proxy Guided Adversarial Defense Framework with Self Distillation Regularization

LoFT: Local Proxy Fine-tuning For Improving Transferability Of Adversarial Attacks Against Large Language Model

Learning Stabilization Control from Observations by Learning Lyapunov-like Proxy Models

Weak Proxies are Sufficient and Preferable for Fairness with Missing Sensitive Attributes

Enhance transferability of adversarial examples with model architecture

Learning Synthetic Environments and Reward Networks for Reinforcement Learning

Decentralized Federated Learning through Proxy Model Sharing