Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

Strong regularization

Strong regularization refers to the application of heavy constraints or large penalty terms during machine learning model training to aggressively restrict model complexity and suppress overfitting. By placing a high coefficient on penalty terms—such as substantial weight decay, high dropout rates, or intensive data perturbations—this approach strictly limits parameter magnitudes and discourages the learning algorithm from excessively relying on specific shortcut features, dominant pathways, or noisy patterns in the training data. While applying strong regularization encourages the model to learn simpler, more robust, and better-generalized representations across all available inputs, an overly aggressive penalty risks underfitting by excessively restricting the capacity of the model to capture meaningful relationships.

1 item

Characterizing and Overcoming the Greedy Nature of Learning in Multi-modal Deep Neural Networks

Characterizing and Overcoming the Greedy Nature of Learning in Multi-modal Deep Neural Networks

Nan Wu, Stanislaw Jastrzebski, Kyunghyun Cho, Krzysztof J. Geras

OrganizationsCIFARGenentechNew York University

Why you should read this

Explains why multi-modal neural networks often over-rely on a single modality and introduces a training algorithm that balances learning speeds across modalities to improve overall generalization.

We hypothesize that due to the greedy nature of learning in multi-modal deep neural networks, these models tend to rely on just one modality while under-fitting the other modalities. Such behavior is counter-intuitive and hurts the models’ generalization, as we observe empirically. To estimate the model’s dependence on each modality, we compute the gain on the accuracy when the model has access to it in addition to another modality. We refer to this gain as the conditional utilization rate. In the experiments, we consistently observe an imbalance in conditional utilization rates between modalities, across multiple tasks and architectures. Since conditional utilization rate cannot be computed efficiently during training, we introduce a proxy for it based on the pace at which the model learns from each modality, which we refer to as the conditional learning speed. We propose an algorithm to balance the conditional learning speeds between modalities during training and demonstrate that it indeed addresses the issue of greedy learning.1 The proposed algorithm improves the model’s generalization on three datasets: Colored MNIST, ModelNet40, and NVIDIA Dynamic Hand Gesture.

Added

2026-09-26