# Deep learning languages: a key fundamental shift from probabilities to weights?

## TL;DR

This paper discusses a fundamental shift in language modeling from probabilistic to weighted representations driven by deep learning successes, highlighting implications for protein sequence classification and the development of non-probabilistic models.

## Contribution

It analyzes the significance of moving from probabilistic to weighted models in language processing, emphasizing the need for principled non-probabilistic learning methods.

## Key findings

- Deep learning enables a shift to weighted representations in language models.
- Probabilistic models face limitations in protein sequence classification.
- Non-probabilistic models require more principled learning approaches.

## Abstract

Recent successes in language modeling, notably with deep learning methods, coincide with a shift from probabilistic to weighted representations. We raise here the question of the importance of this evolution, in the light of the practical limitations of a classical and simple probabilistic modeling approach for the classification of protein sequences and in relation to the need for principled methods to learn non-probabilistic models.