Posters

Researchers dissected attention paths in Transformers using statistical mechanics, revealing a task-relevant kernel combination mechanism boosting generalization performance.

Cross-environment Intra-class Alignment (CIA) and its label-free variant, CIA-LRA, significantly improve node-level OOD generalization on graphs by aligning representations and eliminating spurious fe…

SCD-Tree & GBD: Unlocking interpretable rules for unsupervised anomaly detection!

DISP-LLM: A novel dimension-independent structural pruning method for LLMs achieves accuracy similar to semi-structural pruning while improving flexibility and efficiency, outperforming state-of-the-a…

DISC, a novel control policy, disentangles untrusted ML predictions to achieve near-optimal performance when accurate, while guaranteeing competitive ratio bounds even with significant prediction erro…

Supervised Independent Subspace PCA (sisPCA) disentangles interpretable factors in high-dimensional data by leveraging supervision to maximize subspace dependence on target variables while minimizing …

This study reveals that high input similarity paired with low output similarity is detrimental to continual learning, whereas the opposite scenario is relatively benign; offering insights into mitigat…

DUSDi: A novel method for learning disentangled skills in unsupervised reinforcement learning, enabling efficient reuse for diverse downstream tasks.

This paper introduces a novel implicit Zero-Watermarking scheme using disentangled style domains to detect unauthorized dataset usage in text-to-image models, offering robust copyright protection via …

This paper introduces graphical criteria and an algorithm for disentangling causal factors from heterogeneous data in non-Markovian settings, advancing causal representation learning.

DisenGCD, a meta multigraph framework, disentangles graph learning for cognitive diagnosis, achieving robust student knowledge assessment.

Researchers surpass the 1/e barrier in submodular maximization with novel combinatorial algorithms!

DISCO: a novel discrete-state continuous-time diffusion model for flexible and efficient graph generation, outperforming state-of-the-art methods.

Bridging the gap between continuous diffusion models and discrete data, this work introduces a novel boundary-conditional approach achieving superior performance in language modeling and image generat…

Boosting structured representation learning, a novel Discrete Dictionary-based Decomposition (D3) layer significantly improves systematic generalization in TPR-based models by efficiently decomposing …

COAT leverages LLMs to identify high-level causal factors from unstructured data, enabling causal discovery in real-world scenarios where well-defined variables are lacking.

DSA, a novel automated framework, discovers optimal sparsity allocation for layer-wise LLM pruning, achieving significant performance gains across various models and tasks.

LLMs discover novel offline preference optimization algorithms, achieving state-of-the-art performance on various tasks.

AI discovers robust, biologically-plausible plasticity rules that self-organize and maintain neural circuits’ sequential activity, even with synaptic turnover.

DUPLEX: a novel RL method trains diverse, near-optimal policies in complex, dynamic environments by explicitly maximizing policy diversity using successor features. It outperforms existing methods in…