> ML_ALGORITHM // ENERGY-BASED-MODELS_v1.0

Energy-Based Models (EBM)

Flexible probability modeling framework that parameterizes data distributions via an unnormalized scalar energy function without computing the partition function.

Unnormalized Probability Modelsdeep-generativemoderate-posthoclarge (>100k)

Back to All Algorithms

Computational Complexity

Training Complexity:O(epochs * batch_size * (forward + langevin_steps))

Inference Complexity:O(langevin_mcmc_steps)

Hardware Profile

CPU Friendly:No

Requires GPU:Yes

Memory Footprint:high

Interpretability & Data

Interpretability Tier:moderate-posthoc

Training Data Needs:large (>100k)

Interpretability Assessment

Energy function directly acts as an unnormalized negative log-probability and out-of-distribution scoring metric.

Suitable Tasks & Supported Modalities

Suitable Tasks:

density estimationout of distribution detection

Supported Modalities:

imagetabular

Implementing Libraries

PyTorchLinux Foundation / PyTorch Foundation · v2.4.1

View Spec

Foundational Literature

Common Pitfalls & Warnings

Langevin dynamics MCMC chains fail to mix across isolated energy wells, causing training gradient collapse