baller2vec++: A Look-Ahead Multi-Entity Transformer For Modeling Coordinated Agents (2021-04-24T00:00:00.000000Z)

Abstract

In many multi-agent spatiotemporal systems, agents operate under the influence of shared, unobserved variables (e.g., the play a team is executing in a game of basketball). As a result, the trajectories of the agents are often statistically dependent at any given time step; however, almost universally, multi-agent models implicitly assume the agents' trajectories are statistically independent at each time step. In this paper, we introduce baller2vec++, a multi-entity Transformer that can effectively model coordinated agents. Specifically, baller2vec++ applies a specially designed self-attention mask to a mixture of location and"look-ahead"trajectory sequences to learn the distributions of statistically dependent agent trajectories. We show that, unlike baller2vec (baller2vec++'s predecessor), baller2vec++ can learn to emulate the behavior of perfectly coordinated agents in a simulated toy dataset. Additionally, when modeling the trajectories of professional basketball players, baller2vec++ outperforms baller2vec by a wide margin.

Authors

A. Nguyen

5 papers

Michael A. Alcorn

1 papers

baller2vec++: A Look-Ahead Multi-Entity Transformer For Modeling Coordinated Agents

Abstract

Authors

References27 items

Is Space-Time Attention All You Need for Video Understanding?

baller2vec: A Multi-Entity Transformer For Multi-Agent Spatiotemporal Modeling

Big Bird: Transformers for Longer Sequences

Temporal Graph Networks for Deep Learning on Dynamic Graphs

Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Prediction

Longformer: The Long-Document Transformer

Reformer: The Efficient Transformer

Axial Attention in Multidimensional Transformers

Predicting Dynamic Embedding Trajectory in Temporal Interaction Networks

Social-BiGAT: Multimodal Trajectory Forecasting using Bicycle-GAN and Graph Attention Networks

XLNet: Generalized Autoregressive Pretraining for Language Understanding

Argoverse: 3D Tracking and Forecasting With Rich Maps

Diverse Generation for Multi-Agent Sports Games

Human motion trajectory prediction: a survey

Generating Long Sequences with Sparse Transformers

Multi-Agent Tensor Fusion for Contextual Trajectory Prediction

TraPHic: Trajectory Prediction in Dense and Heterogeneous Traffic Using Weighted Interactions

Where Will They Go? Predicting Fine-Grained Adversarial Multi-agent Motion Using Conditional Variational Autoencoders

SoPhie: An Attentive GAN for Predicting Paths Compliant to Social and Physical Constraints

Convolutional Social Pooling for Vehicle Trajectory Prediction

Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networks

Generating Multi-Agent Trajectories using Programmatic Weak Supervision

Graph Attention Networks

Attention is All you Need

Generating Long-term Trajectories Using Deep Hierarchical Networks

Adam: A Method for Stochastic Optimization

(batter|pitcher)2vec : Statistic-Free Talent Modeling With Neural Player Embeddings

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names