Welcome to MultiBench’s Documentation!

Welcome to the documentation for MultiBench/MultiZoo, a comprehensive evaluation and set of implementations of existing multimodal machine learning algorithms. It contains 20 methods spanning different methodological innovations in (1) data preprocessing, (2) fusion paradigms, (3) optimization objectives, and (4) training procedures.

---

Getting Started

• Installation
  • Prerequisites
  • Virtual environment
  • Installing dependencies
• Quick Start
  • A first experiment: early fusion on CMU-MOSI
  • Quickest experiments to get started
  • Smoke-test results
  • Running other experiments
• Downloading Datasets
  • Affective Computing
  • Healthcare
  • Robotics
  • Finance
  • HCI
  • Multimedia
• Continuous Integration Guide
  • Repo Structure
  • Dataloading, Training, and Evaluation
  • Testing and Contributing

Tutorials:

• Simple Early Fusion Case
• Multimodal Fusion Architecture Search
• MCTN

Contents:

• Datasets and DataLoaders
  • datasets package
• General Evaluation Scripts
  • eval_scripts package
• Multimodal Fusion Techniques
  • fusions package
• Objective Functions
  • objective_functions package
• Robustness
  • robustness package
• Training Structures
  • training_structures package
• Unimodal Encoders
  • unimodals package
• Utilities
  • utils package

Indices and tables

• Index

• Module Index

• Search Page