4 papers
Conservation Law Breaking at the Edge of Stability: A Spectral Theory of Non-Convex Neural Network Optimization
Daniel Nobrega Medeiros
Why does gradient descent reliably find good solutions in non-convex neural network optimization, despite the landscape being NP-hard in the worst case? We show that gradient flow…
I Can't Believe TTA Is Not Better: When Test-Time Augmentation Hurts Medical Image Classification
Daniel Nobrega Medeiros
Test-time augmentation (TTA)--aggregating predictions over multiple augmented copies of a test input--is widely assumed to improve classification accuracy, particularly in medical…
PolyGLU: State-Conditional Activation Routing in Transformer Feed-Forward Networks
Daniel Nobrega Medeiros
Biological neural systems employ diverse neurotransmitters -- glutamate, GABA, dopamine, acetylcholine -- to implement distinct signal-processing modalities within shared neural ci…
TACIT Benchmark: A Programmatic Visual Reasoning Benchmark for Generative and Discriminative Models
Daniel Nobrega Medeiros
Existing visual reasoning benchmarks predominantly rely on natural language prompts, evaluate narrow reasoning modalities, or depend on subjective scoring procedures such as LLM-as…