What we are reading: paper pills
Paper pills are summaries of publications, talks or new software, with context and a short description of the methods, main ideas and key results. They are written for practitioners or researchers who want a peek into recent developments but are short on time. We publish monthly digests as well as highlights of major conferences.

Pills

A Trust Crisis In Simulation-Based Inference? Your Posterior Approximations Can Be Unfaithful

In posterior estimation, it is generally worse to exclude plausible parameters than wrongly including implausible ones. However, extensive tests yield empirical evidence that popular algorithms provide overconfident estimations.

A picture of Maternus
Simulation-Based Inference

Benchmarking Simulation-Based Inference

A much-needed benchmark for methods of simulation-based inference.

A picture of Maternus
Simulation-Based Inference

The Frontier of Simulation-based Inference

An overview and schematic comparison of recent developments in simulation-based inference and their enabling factors. Advancements in ML, Active Learning and Augmentation are named as the three driving forces in the field.

A picture of Maternus
Simulation-Based Inference

Beyond neural scaling laws: beating power law scaling via data pruning

Large neural networks are very “data hungry”, but data pruning holds the promise to alleviate this dependence. A recent paper studies the use of different pruning strategies and metrics, with encouraging results.

A photo of Fabio
Data Valuation

What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation

When data samples are difficult to learn, neural networks tend to memorize their labels rather than infer useful features. This has been shown to improve, rather than limit, their accuracy. A recent paper introduces a few key concepts that help investigate this phenomenon.

A photo of Fabio
Data Valuation

Representation learning with BYOL and SimSiam

BYOL was the first work to show how useful low-dimensional representations can be when learned in an unsupervised way without negative sampling. It inspired a series of simpler architectures, with SimSiam among them.

A picture of Michael
Representation Learning

CS-Shapley: Class-wise Shapley Values for Data Valuation in Classification

Using in-class accuracy to up-weight the value of a data point and out-of-class accuracy as a discounting factor, the authors define a new utility function that is better suited for valuation in classification tasks.

A picture of Miguel
Data Valuation

Mixture of Decision Trees for Interpretable Machine Learning

A linear gating function together with multiple expert decision trees is trained by expectation-maximization and results in a new, fully interpretable model that works well on several (simple) data sets.

A picture of Michael
Explainable AI

Critic Regularized Regression

A simple but powerful algorithm for offline reinforcement learning, which can be seen as a combination of behavior cloning and Q-learning, and sets a new state-of-the-art in many tasks.

A picture of Michael
Efficient RL

Data Banzhaf: A Robust Data Valuation Framework for Machine Learning

A definition of data value with much higher rank stability than Shapley values, proven to be most robust among all semi-values, mitigates the pressing issue of utility variance in ML applications.

A picture of Miguel
Data Valuation