paper-with-me

홈 › Papers

Should We Learn Most Likely Functions or Parameters?

2023-11-27 · NeurIPS 2023 11 · Shikai Qiu, Tim G. J. Rudner, Sanyam Kapoor, Andrew Gordon Wilson

Standard regularized training procedures correspond to maximizing a posterior distribution over parameters, known as maximum a posteriori (MAP) estimation. However, model parameters are of interest only insomuch as they combine with the functional form of a model to provide a function that can make good predictions. Moreover, the most likely parameters under the parameter posterior do not generally correspond to the most likely function induced by the parameter posterior. In fact, we can re-parametrize a model such that any setting of parameters can maximize the parameter posterior. As an alternative, we investigate the benefits and drawbacks of directly estimating the most likely function implied by the model and the data. We show that this procedure leads to pathological solutions when using neural networks and prove conditions under which the procedure is well-behaved, as well as a scalable approximation. Under these conditions, we find that function-space MAP estimation can lead to flatter minima, better generalization, and improved robustness to overfitting.

📄 PDF Abstract BibTeX arXiv:2311.15990

Code (1)

activatedgeek/function-space-map 공식 구현 jax

Similar Papers 제목 키워드 기반

It Is Likely That Your Loss Should be a Likelihood

2020-07-12 · Mark Hamilton, Evan Shelhamer, William T. Freeman

Many common loss functions such as mean-squared-error, cross-entropy, and reconstruction loss are unnecessarily rigid. Under a probabilistic interpretation, these common losses correspond to distributions with fixed shap…

Outlier Detection

An ideal observer model for identifying the reference frame of objects

2011-12-01 · NeurIPS 2011 12 · Joseph L. Austerweil, Abram L. Friesen, Thomas L. Griffiths

The object people perceive in an image can depend on its orientation relative to the scene it is in (its reference frame). For example, the images of the symbols $\times$ and $+$ differ by a 45 degree rotation. Although …

Decoupling Learning Rules from Representations

2017-06-09 · Philip S. Thomas, Christoph Dann, Emma Brunskill

In the artificial intelligence field, learning often corresponds to changing the parameters of a parameterized function. A learning rule is an algorithm or mathematical expression that specifies precisely how the paramet…

Reinforcement Learning

Decoupling Gradient-Like Learning Rules from Representations

2018-07-01 · ICML 2018 7 · Philip Thomas, Christoph Dann, Emma Brunskill

In machine learning, learning often corresponds to changing the parameters of a parameterized function. A learning rule is an algorithm or mathematical expression that specifies precisely how the parameters should b…

BIG-bench Machine LearningReinforcement Learning

Adaptive Pairwise Weights for Temporal Credit Assignment

2021-02-09 · Zeyu Zheng, Risto Vuorio, Richard Lewis, Satinder Singh

How much credit (or blame) should an action taken in a state get for a future reward? This is the fundamental temporal credit assignment problem in Reinforcement Learning (RL). One of the earliest and still most widely u…

Reinforcement Learning (RL)