Trending

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

Machine Learning

Stats Machine Learning

Marginal-Likelihood Based Model Selection in Deep Learning

Scalable Marginal Likelihood Estimation for Model Selection in Deep Learning

Marginal-likelihood based model-selection, even though promising, is rarely used in deep learning due to estimation difficulties.

Instead, most approaches rely on validation data, which may not be readily available.

In this work, we present a scalable marginal-likelihood estimation method to select both the hyperparameters and network architecture based on the training data alone, and it outperforms cross-validationand manual-tuning on standard regression and image classification datasets, especially in terms of calibration and out-of-distribution detection.

Authors

Alexander Immer, Matthias Bauer, Vincent Fortuin, Gunnar Rätsch, Mohammad Emtiyaz Khan

Related Topics

Network architecture

Cross-validation

Laplace's method

Marginal-likelihood based model-selection

Scalable marginal-likelihood estimation method

Gauss-newton approximations

Marginal-likelihood estimate

Read the Paper

Read Paper

Trending

Bookmarks

◐ Recommended Follows

Mohammadreza Rezaee

More

◐ Latest Activity

More

Marginal-Likelihood Based Model Selection in Deep Learning - 42Papers

Read Related Papers

Multi-Scale Deep Learning for Estimating Horizontal Velocity Fields on the Solar Surface

Criticality of Deep Neural Networks with Infinity Approximation

Multiset-Equivalence in Deep Set Prediction

Data Selection Based Fine-tuning Pipeline for Image Classification

Adversarial AutoAugment

Softmax Gradient Tampering: Decoupling the Backward Pass for Improved Fitting

A Method for Evaluating the Capacity of Generative Adversarial Networks to Reproduce High-order Spatial Context

Learning to Refit for Convex Learning Problems

Disparity Between Batches as a Signal for Early Stopping

Contrastive Deep Learning for Mammography Lesion Detection