Early stopping by correlating online indicators in neural networks
This work addresses the challenge of overfitting in neural network training for practitioners, offering a more reliable early stopping method, though it appears incremental as it builds on existing overfitting identification techniques.
The paper tackles the problem of minimizing generalization error in neural networks by introducing a novel technique that identifies overfitting through correlation of multiple online indicators, enabling reliable early stopping. In a case study on parser generation for natural language processing, the method shows promising results compared to existing single-criterion approaches.
In order to minimize the generalization error in neural networks, a novel technique to identify overfitting phenomena when training the learner is formally introduced. This enables support of a reliable and trustworthy early stopping condition, thus improving the predictive power of that type of modeling. Our proposal exploits the correlation over time in a collection of online indicators, namely characteristic functions for indicating if a set of hypotheses are met, associated with a range of independent stopping conditions built from a canary judgment to evaluate the presence of overfitting. That way, we provide a formal basis for decision making in terms of interrupting the learning process. As opposed to previous approaches focused on a single criterion, we take advantage of subsidiarities between independent assessments, thus seeking both a wider operating range and greater diagnostic reliability. With a view to illustrating the effectiveness of the halting condition described, we choose to work in the sphere of natural language processing, an operational continuum increasingly based on machine learning. As a case study, we focus on parser generation, one of the most demanding and complex tasks in the domain. The selection of cross-validation as a canary function enables an actual comparison with the most representative early stopping conditions based on overfitting identification, pointing to a promising start toward an optimal bias and variance control.