Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AI for Handball: predicting and explaining the 2024 Olympic Games tournament with Deep Learning and Large Language Models

Florian Felice

TL;DR

This work tackles the challenge of predicting and explaining the 2024 Paris Olympic handball tournament by integrating a multi-modal deep learning model with explainable AI and a Large Language Model for human-friendly explanations. The model uses lineup embeddings and transfer learning from club data to better capture team composition, while Integrated Gradients provide input-attribution scores that feed into an LLM (Mistral-7B-Instruct) to generate coach-ready summaries. Results show that transfer learning reduces prediction error on international data, with explicit predictions for Paris 2024 that align with perceived favorites across both genders; explanations illustrate concrete drivers such as lineup cohesion, backline strength, and international experience. The framework demonstrates a practical, coach-oriented AI pipeline that links precise predictions with actionable narratives, though data availability and computational considerations remain key limitations for broader adoption.

Abstract

Over summer 2024, the world will be looking at Paris to encourage their favorite athletes win the Olympic gold medal. In handball, few nations will fight hard to win the precious metal with speculations predicting the victory for France or Denmark for men and France or Norway for women. However, there is so far no scientific method proposed to predict the final results of the competition. In this work, we leverage a deep learning model to predict the results of the handball tournament of the 2024 Olympic Games. This model, coupled with explainable AI (xAI) techniques, allows us to extract insightful information about the main factors influencing the outcome of each match. Notably, xAI helps sports experts understand how factors like match information or individual athlete performance contribute to the predictions. Furthermore, we integrate Large Language Models (LLMs) to generate human-friendly explanations that highlight the most important factors impacting the match results. By providing human-centric explanations, our approach offers a deeper understanding of the AI predictions, making them more actionable for coaches and analysts.

AI for Handball: predicting and explaining the 2024 Olympic Games tournament with Deep Learning and Large Language Models

TL;DR

This work tackles the challenge of predicting and explaining the 2024 Paris Olympic handball tournament by integrating a multi-modal deep learning model with explainable AI and a Large Language Model for human-friendly explanations. The model uses lineup embeddings and transfer learning from club data to better capture team composition, while Integrated Gradients provide input-attribution scores that feed into an LLM (Mistral-7B-Instruct) to generate coach-ready summaries. Results show that transfer learning reduces prediction error on international data, with explicit predictions for Paris 2024 that align with perceived favorites across both genders; explanations illustrate concrete drivers such as lineup cohesion, backline strength, and international experience. The framework demonstrates a practical, coach-oriented AI pipeline that links precise predictions with actionable narratives, though data availability and computational considerations remain key limitations for broader adoption.

Abstract

Over summer 2024, the world will be looking at Paris to encourage their favorite athletes win the Olympic gold medal. In handball, few nations will fight hard to win the precious metal with speculations predicting the victory for France or Denmark for men and France or Norway for women. However, there is so far no scientific method proposed to predict the final results of the competition. In this work, we leverage a deep learning model to predict the results of the handball tournament of the 2024 Olympic Games. This model, coupled with explainable AI (xAI) techniques, allows us to extract insightful information about the main factors influencing the outcome of each match. Notably, xAI helps sports experts understand how factors like match information or individual athlete performance contribute to the predictions. Furthermore, we integrate Large Language Models (LLMs) to generate human-friendly explanations that highlight the most important factors impacting the match results. By providing human-centric explanations, our approach offers a deeper understanding of the AI predictions, making them more actionable for coaches and analysts.
Paper Structure (28 sections, 5 figures, 6 tables)

This paper contains 28 sections, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Motivation
  3. Literature review
  4. Materials and methods
  5. Data
  6. Match information
  7. Teams information
  8. Teams strengths
  9. Teams lineups
  10. Neural networks and transfer learning
  11. Embedding Multi-Layered Perceptron
  12. Transfer Learning
  13. Large Language Model for human readable explanations
  14. LLM inputs
  15. Few-shot prompting
  16. ...and 13 more sections

Figures (5)

  • Figure 1: Architecture of the multi-modal neural network. The textual inputs are passed to an embedding layer before being merged with the numerical inputs to later be fed to the fully connected part of the network.
  • Figure 2: Evolution of the training and validation loss over epochs with and without transfer learning (on men national teams data). The loss used is the sum of Mean Squared Error for home and away teams.
  • Figure 3: Predictions for the final round of the 2024 Olympic Games for the men's tournament
  • Figure 4: Predictions for the final round of the 2024 Olympic Games for the women's tournament
  • Figure 5: Local explanation of the final match France vs Croatia of the men's tournament. Attribution values are calculated using Integrated Gradients from the deep neural network as presented in Section \ref{['sec:model']}.