VAIN: Attentional Multi-agent Predictive Modeling
By Yedid Hoshen. 2017
255 citations recorded by the source corpus.
Multi-agent predictive modeling is an essential step for understanding physical, social and team-play systems. Recently, Interaction Networks (INs) were proposed for the task of modeling multi-agent physical systems, INs scale with the number of interactions in the system (typically quadratic or higher order in the number of agents). In this paper we introduce VAIN, a novel attentional architecture for multi-agent predictive modeling that scales linearly with the number of agents. We show that VAIN is effective for multi-agent predictive modeling. Our method is evaluated on tasks from challenging multi-agent prediction domains: chess and soccer, and outperforms competing multi-agent approaches.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, LLM Agents & Tool Use, Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yoshua Bengio
- Sergey Levine
- Mohit Bansal
- Jianfeng Gao
- Quoc V. Le
- Percy Liang
- Ruslan Salakhutdinov
- Kyunghyun Cho
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Trevor Darrell
- Xiaogang Wang
- Ming-Hsuan Yang
- Andrew Zisserman
- Jitendra Malik
- Andrea Vedaldi
In LLM Agents & Tool Use
- Davood Hajinezhad
- Afshin Oroojlooyjadid
- Joelle Pineau
- Tonghan Wang
- Dhruv Batra
- Zongqing Lu
- Jiechuan Jiang
- Jianhao Wang
In Signal Processing
- Yoshua Bengio
- J. Urgen Schmidhuber
- Li Deng
- Dong Yu
- Oriol Vinyals
- Chris Dyer
- Max Welling
- Ruslan Salakhutdinov
Citations in the corpus
- Graph Attention Networks (2017, background): “Moreover, our approach of sharing a neural network computation across edges is reminiscent of the formulation of relational networks (Santoro et al., 2017) and VAIN (Hoshen, 2017), wherein relations between objects or agents are aggregated pair-wise, by employing a shared mechanism.”
- Relation Networks for Object Detection (2017, methodology): “Attention modules in NLP and physical system modeling Attention modules have recently been successfully applied in the NLP field [20, 21, 6, 56] and in physical system modeling [2, 57, 31, 45, 13, 42].”
- Stochastic Prediction of Multi-Agent Interactions from Partial Observations (2019, positive): “It is also possible to use graph attention networks (Veli ˇ ckovi ´ c et al., 2018) for this task (see e.g., Hoshen (2017); Battaglia et al. (2018)).”
- MONet: Unsupervised Scene Decomposition and Representation (2019, background): “…networks that rely on hand-crafted object representations have recently achieved remarkable results in a wide range of research areas, including reinforcement learning, physical modeling, and multi-agent control [Battaglia et al., 2018, Wang et al., 2018, Hamrick et al., 2017, Hoshen, 2017].”
- Relational inductive biases, deep learning, and graph networks (2018, methodology): “…vein, Wang et al. (2018c) introduced the non-local neural network (NLNN), which unified various “self-attention”-style methods (Vaswani et al., 2017; Hoshen, 2017; Veliˇckovi´c et al., 2018) by analogy to methods from computer vision and graphical models for capturing long range dependencies in…”
- SE(3)-Transformers: 3D Roto-Translation Equivariant Attention Networks (2020, background): “[32] and Hoshen [12].”
- Attention-based Graph Neural Network for Semi-supervised Learning (2018, background): “Here we note that independently from this work attention over sets has been proposed as “neighborhood attention” [16, 25] for a different application.”
- Learning Nearly Decomposable Value Functions Via Communication Minimization (2019, background): “n et al., 2019). Communication is a hot topic in multi-agent reinforcement learning. End-to-end learning with differentiable communication channel is a popular approach now. Sukhbaatar et al. (2016); Hoshen (2017); Jiang & Lu (2018); Singh et al. (2019); Das et al. (2019) focus on learning decen”
- Relational Forward Models for Multi-Agent Learning (2018, positive): “First, we show that our models can surpass previous top methods on this task (Kipf et al., 2018; Hoshen, 2017), and, crucially, produce intermediate representations that support the social analysis of multi-agent systems: our models explain what drives each agent’s behavior, track how agents…”
- Analogues of mental simulation and imagination in deep learning (2019, unknown): “Recently, models which represent the state as a graph [34 ] have been used to predict the motion of rigid bodies [35,36], deformable objects [37], articulated robotic systems [38], and multiagent systems [39].”
- Deep Learning on Graphs: A Survey (2018, unknown): “…[137], [138], [139], gene expression [140], natural language processing (NLP) [141], [142], computer vision [143], [144], [145], [146], [147], traffic forecasting [148], [149], program induction [150], solving graph-based NP problems [151], [152], and multi-agent AI systems [153], [154], [155].”
- Learning Transferable Cooperative Behavior in Multi-Agent Teams (2019, methodology): “VAIN (Hoshen, 2017) improves upon the mean aggregation by using an exponential kernel based attention to selectively attend to the messages received from other agents, and showed predictive modeling of multi-agent systems using supervised learning.”
- Learning Distilled Collaboration Graph for Multi-Agent Perception (2021, methodology): “For example, CommNet [32] adopted the averaging operation to achieve continuous communication in multi-agent system; VAIN [11] considered an attention mechanism to determine which agents would share information;”
- Concurrent Meta Reinforcement Learning (2019, unknown): “nts that was either discrete or continuous, with potential channel noise added. These early architectures showed success on a variety of simple environments such as a traffic junction simulation. VAIN Hoshen (2017) extended CommNet with an attention mechanism that allowed agents to focus on the featu”
- Graph Convolutional Reinforcement Learning (2018, background): “A few interaction frameworks have been proposed to predict the future states and underlying properties, such as IN (Battaglia et al., 2016), VIN (Watters et al., 2017), and VAIN (Hoshen, 2017).”
- VectorNet: Encoding HD Maps and Agent Dynamics From Vectorized Representation (2020, background): “[18] treats such interactions as latent variables; and graph attention networks [16, 31] apply self-attention mechanism to weight the edges in a pre-defined graph.”
- Learning Policy Representations in Multiagent Systems (2018, background): “Graphs are a fundamental abstraction for modeling relational data, such as the interactions arising in multiagent systems (Zhou et al., 2016a;b; Chen et al., 2017; Battaglia et al., 2016; Hoshen, 2017) and concurrent work proposes to learn such graphs directly from data (Kipf et al.”