# Tensor-variate Mixture of Experts for Proportional Myographic Control of a Robotic Hand

## TL;DR

This paper introduces a tensor-based mixture-of-experts model for regression tasks involving tensor data, effectively capturing data structure and reducing overfitting, demonstrated through myography-based hand movement recognition.

## Contribution

The paper proposes a novel tensor-variate mixture-of-experts model that leverages tensor structures for improved regression with limited data.

## Key findings

- Effective in recognizing hand movements from tactile myography
- Performs well with limited training data
- Outperforms traditional flattening-based methods

## Abstract

When data are organized in matrices or arrays of higher dimensions (tensors), classical regression methods first transform these data into vectors, therefore ignoring the underlying structure of the data and increasing the dimensionality of the problem. This flattening operation typically leads to overfitting when only few training data is available. In this paper, we present a mixture-of-experts model that exploits tensorial representations for regression of tensor-valued data. The proposed formulation takes into account the underlying structure of the data and remains efficient when few training data are available. Evaluation on artificially generated data, as well as offline and real-time experiments recognizing hand movements from tactile myography prove the effectiveness of the proposed approach.