# On Expected Accuracy

## TL;DR

This paper explores an alternative loss function, expected accuracy, for classification tasks, demonstrating its comparable or superior performance and robustness to label noise across various neural network architectures.

## Contribution

It introduces a modified, leaky version of expected accuracy as a loss function and evaluates its effectiveness across multiple tasks and models.

## Key findings

- Expected accuracy achieves comparable or better accuracy than cross entropy.
- The leaky expected accuracy is more robust to label noise.
- Applicable across diverse neural architectures and tasks.

## Abstract

We empirically investigate the (negative) expected accuracy as an alternative loss function to cross entropy (negative log likelihood) for classification tasks. Coupled with softmax activation, it has small derivatives over most of its domain, and is therefore hard to optimize. A modified, leaky version is evaluated on a variety of classification tasks, including digit recognition, image classification, sequence tagging and tree tagging, using a variety of neural architectures such as logistic regression, multilayer perceptron, CNN, LSTM and Tree-LSTM. We show that it yields comparable or better accuracy compared to cross entropy. Furthermore, the proposed objective is shown to be more robust to label noise.