PolyStep optimizes non-differentiable networks via forward-only polytope evaluations and optimal-transport barycentric updates, reaching 93.4% accuracy on hard-LIF spiking networks while outperforming gradient-free baselines.
Pearlmutter, Don Syme, Frank Wood, and Philip Torr
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
years
2026 8verdicts
UNVERDICTED 8representative citing papers
The Program Hypergraph extends binary program semantic graphs to arbitrary-arity hyperedges to faithfully represent multi-way relations in geometric algebra and spatial architectures.
A dimensional type system extends Hindley-Milner inference with abelian-group constraints and carries annotations through MLIR lowering to jointly decide numeric representation and deterministic memory allocation at compile time.
Develops randomized-subspace Nesterov accelerated gradient methods with accelerated oracle-complexity guarantees for smooth convex optimization under matrix smoothness and sketch moment assumptions.
Zeroth-order optimization is underexplored rather than underpowered in deep learning, with limitations stemming from full-space designs that can be addressed via subspace, spectral, and systems-aware approaches.
Echo Networks are recurrent networks defined by a single connection matrix with no layers, enabling matrix-based mutation and recombination in neuroevolution, and demonstrated on ECG signal classification.
A type system over finitely generated abelian groups enables design-time verification of AI model properties and links Hindley-Milner unification to a restriction of Solomonoff's universal prior.
The paper claims that composing the Dimensional Type System, Program Hypergraph, and b-posit 2026 standard yields depth-independent training memory at ~2x inference, grade-preserving updates, Bayesian distillation for domain adaptation, and warm rotation for uninterrupted deployment.
citing papers explorer
-
Training Non-Differentiable Networks via Optimal Transport
PolyStep optimizes non-differentiable networks via forward-only polytope evaluations and optimal-transport barycentric updates, reaching 93.4% accuracy on hard-LIF spiking networks while outperforming gradient-free baselines.
-
The Program Hypergraph: Multi-Way Relational Structure for Geometric Algebra, Spatial Compute, and Physics-Aware Compilation
The Program Hypergraph extends binary program semantic graphs to arbitrary-arity hyperedges to faithfully represent multi-way relations in geometric algebra and spatial architectures.
-
Dimensional Type Systems and Deterministic Memory Management: Design-Time Semantic Preservation in Native Compilation
A dimensional type system extends Hindley-Milner inference with abelian-group constraints and carries annotations through MLIR lowering to jointly decide numeric representation and deterministic memory allocation at compile time.
-
Randomized Subspace Nesterov Accelerated Gradient
Develops randomized-subspace Nesterov accelerated gradient methods with accelerated oracle-complexity guarantees for smooth convex optimization under matrix smoothness and sketch moment assumptions.
-
Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered
Zeroth-order optimization is underexplored rather than underpowered in deep learning, with limitations stemming from full-space designs that can be addressed via subspace, spectral, and systems-aware approaches.
-
Introducing Echo Networks for Computational Neuroevolution
Echo Networks are recurrent networks defined by a single connection matrix with no layers, enabling matrix-based mutation and recombination in neuroevolution, and demonstrated on ECG signal classification.
-
Decidable By Construction: Design-Time Verification for Trustworthy AI
A type system over finitely generated abelian groups enables design-time verification of AI model properties and links Hindley-Milner unification to a restriction of Solomonoff's universal prior.
-
Adaptive Domain Models: Bayesian Evolution, Warm Rotation, and Principled Training for Geometric and Neuromorphic AI
The paper claims that composing the Dimensional Type System, Program Hypergraph, and b-posit 2026 standard yields depth-independent training memory at ~2x inference, grade-preserving updates, Bayesian distillation for domain adaptation, and warm rotation for uninterrupted deployment.