We construct a Neural Network that approximates matrix multiplication operator for any activation function such that there exists a Neural Network which can approximate the scalar multiplication function. In particular, we use the Strassen algorithm to bound the number of weights and layers needed for such Neural Networks. This allows us to define another Neural Network for approximating the inverse matrix operator. Finally, we discuss improvements with respect to the prior results.

For the full statement of these results see: https://arxiv.org/abs/2501.06539.

This talk is based on a joint work with Jon Asier B\`{a}rcena-P\`{e}tisco.