Conference on Parsimony and Learning (CPAL)

March 2025, Stanford

Subject Areas

Theory & Foundations

• Theories for sparse coding, structured sparsity, subspace learning, low-dimensional manifolds, and general low-dimensional structures.
• Dictionary learning and representation learning for low-dimensional structures and their connections to deep learning theory.
• Equivariance and invariance modeling.
• Theoretical neuroscience and cognitive science foundation for parsimony, and biologically inspired computational mechanisms.

Optimization & Algorithms

• Optimization, robustness, and generalization methods for learning compact and structured representations.
• Interpretable and efficient deep architectures (e.g., based on unrolled optimization).
• Data-efficient and computation-efficient training and inference.
• Adaptive and robust learning and inference algorithms.
• Distributed, networked, or federated learning at scale.
• Other nonlinear dimension-reduction and representation-learning methods.

Data, Systems & Applications

• Domain-specific datasets, benchmarks, and evaluation metrics.
• Parsimonious and structured representation learning from data.
• Inverse problems that benefit from parsimonious priors.
• Hardware and system co-design for parsimonious learning algorithms.
• Parsimonious learning in intelligent systems that integrate perception-action cycles.
• Applications in science, engineering, medicine, and social sciences.

The above is intended as a high-level overview of CPAL’s focus and by no means exclusive. If you doubt that your paper fits the venue, feel free to contact the program chairs via email at pcs@cpal.cc.