Are Easier or Harder Examples Better? Rethinking Data Selection for Reward Models and Preference Optimization

Kevin Christian Wibisono, Aya Ismail, Pedro O. Pinheiro, Yixin Wang and 3 more

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

Active Learning of Conditional Generative Models via the Transport Neural Tangent Kernel

Jayoung Ryu, Kyunghyun Cho, Romain Lopez

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Understanding Reasoning from Pretraining to Post-Training: Chess as a Controlled Testbed

Jingyan Shen, Ang Li, Salman Rahman, Yifan Sun and 3 more

Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Epiplexity Guided Data Selection and Generation for Out-of-Distribution Generalization

Epiplexity guides data selection and synthetic generation to improve out-of-distribution transfer by favoring structurally rich training data.

Ellen Su, Andres Potapczynski, Shikai Qiu, Edward Hughes and 1 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Prospective Coding Improves Learning in Deep Continuous-Time Recurrent Networks

Prospective bottom-up inputs via two-tap updates mitigate depth-dependent gradient attenuation in deep continuous-time recurrent networks, boosting RQF accuracy on Speech Commands and Path-X.

Shivang Rawat, Mirko Morello, Flaviano Morone, David Heeger

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

On-Policy Consistency Training Improves LLM Safety with Minimal Capability Degradation

On-Policy Consistency Training improves LLM safety across sycophancy, jailbreaks, and safety awareness while avoiding the capability regressions of supervised fine-tuning.

Andy Q Han, Kristina Fujimoto, Avidan Shah, Kiet Nguyen and 4 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Uniform-in-Time Weak Propagation of Chaos in Shallow Neural Networks

Shallow neural networks trained via gradient descent exhibit uniform-in-time weak propagation of chaos, yielding poly(d/ε) neuron and sample complexity when mean-field loss decays faster than t^{-2}.

Margalit Glasgow, Joan Bruna

Atlanta Poster Session 5, Fri, Dec 11, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Geometric Factual Recall in Transformers

Transformers memorize facts geometrically via linear superpositions and MLP selectors, needing only logarithmic dimensions and enabling zero-shot MLP transfer.

Shauli Ravfogel, Gilad Yehudai, Joan Bruna, Alberto Bietti

Paris Poster Session 3, Thu, Dec 10, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026 · ▲ 1 on Hugging Face

On the Meta-Design of Allocation Problems

The paper defines meta-design for resource allocation by optimizing upstream design parameters like data, capacity, and quality, and demonstrates the framework in German employment and Ethiopian cash transfer programs.

Unai Fischer Abaigar, Emily Aiken, Christoph Kern, Juan C Perdomo

Atlanta Poster Session 5, Fri, Dec 11, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Explanation Multiplicity in SHAP: Characterization and Assessment

SHAP produces multiple valid yet different explanations for identical predictions due to intrinsic stochasticity, and magnitude-based stability metrics mask substantial rank instability across datasets and models.

Hyunseung Hwang, Seungeun Lee, Lucas Rosenblatt, Steven Whang and 1 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Neural Neural Scaling Laws

NeuNeu predicts downstream scaling via time-series extrapolation of task accuracies and token-level losses, achieving 1.99% MAE and 44% lower error than logistic scaling laws with zero-shot generalization.

Michael Hu, Jane Pan, Ayush Rajesh Jhaveri, Nicholas Lourie and 1 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026