Pith. sign in

REVIEW 1 cited by

Evaluation metrics for behaviour modeling

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2007.12298 v1 pith:AHB4ES2U submitted 2020-07-23 cs.LG stat.ML

Evaluation metrics for behaviour modeling

classification cs.LG stat.ML
keywords behaviormetricsmodelbiasesbiologistscomparecriteriadistribution
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

A primary difficulty with unsupervised discovery of structure in large data sets is a lack of quantitative evaluation criteria. In this work, we propose and investigate several metrics for evaluating and comparing generative models of behavior learned using imitation learning. Compared to the commonly-used model log-likelihood, these criteria look at longer temporal relationships in behavior, are relevant if behavior has some properties that are inherently unpredictable, and highlight biases in the overall distribution of behaviors produced by the model. Pointwise metrics compare real to model-predicted trajectories given true past information. Distribution metrics compare statistics of the model simulating behavior in open loop, and are inspired by how experimental biologists evaluate the effects of manipulations on animal behavior. We show that the proposed metrics correspond with biologists' intuitions about behavior, and allow us to evaluate models, understand their biases, and enable us to propose new research directions.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Agent-Centric Animal Pose Forecasting

    cs.LG 2026-07 conditional novelty 6.0

    Agent-centric transformers trained through a composable library reproduce several marginal statistics of courting fly behavior, but discriminators still separate simulated from real flies.