arXiv subjects

Kevin J. Miller

Publications and source records attributed to Kevin J. Miller.

5 recordsLinked to original sources

arXiv2026

ATLAS: Active Theory Learning for Automated Science

Advancing scientific understanding through mechanistic modeling requires posing the right experimental questions to yield maximally informative data. To automate this pursuit within cognitive science, we introduce ATLAS (Active Theory Learning for Automated Science), an active learning framework for the data-driven discovery of interpretable behavioral models. ATLAS iterates between generating mechanistic hypotheses--instantiated as a diverse ensemble of sparse neural networks (Disentangled RNNs)--and designing experiments that optimally distinguish between them. We test this approach on the problem of recovering reinforcement learning agents from their behavior in bandit tasks. ATLAS designs varied sequences of qualitatively novel experiments with temporal structure tailored to underlying agent characteristics. The models trained on these experiments are evaluated against a comprehensive set of metrics for mechanistic modeling that capture behavioral, structural, and computational similarity. ATLAS achieves a 5-10x improvement in sample efficiency across all metrics compared to random experimentation, and its performance is further validated against expert-designed experiments derived from literature. These in silico results showcase ATLAS's potential to accelerate human-interpretable insights in cognitive science and other domains where scientific inquiry relies on discovering mechanistic models.

cs.LG↗

arXiv2023

Conditional Randomization Tests for Behavioral and Neural Time Series

Randomization tests allow simple and unambiguous tests of null hypotheses, by comparing observed data to a null ensemble in which experimentally-controlled variables are randomly resampled. In behavioral and neuroscience experiments, however, the stimuli presented often depend on the subject's previous actions, so simple randomization tests are not possible. We describe how conditional randomization can be used to perform exact hypothesis tests in this situation, and illustrate it with two examples. We contrast conditional randomization with a related approach of tangent randomization, in which stimuli are resampled based only on events occurring in the past, which is not valid for all choices of test statistic. We discuss how to design experiments that allow conditional randomization tests to be used.

stat.ME↗

arXiv2020

Deep Reinforcement Learning and its Neuroscientific Implications

The emergence of powerful artificial intelligence is defining new research directions in neuroscience. To date, this research has focused largely on deep neural networks trained using supervised learning, in tasks such as image classification. However, there is another area of recent AI work which has so far received less attention from neuroscientists, but which may have profound neuroscientific implications: deep reinforcement learning. Deep RL offers a comprehensive framework for studying the interplay among learning, representation and decision-making, offering to the brain sciences a new set of research tools and a wide range of novel hypotheses. In the present review, we provide a high-level introduction to deep RL, discuss some of its initial applications to neuroscience, and survey its wider implications for research on brain and behavior, concluding with a list of opportunities for next-stage research.

cs.AI↗

arXiv2020

Sub-picosecond all-optical switching in a hybrid VO2:silicon waveguide at 1550 nm

Achieving ultrafast all-optical switching in a silicon waveguide geometry is a key milestone on the way to an integrated platform capable of handling the increasing demands for higher speed and higher capacity for information transfer. Given the weak electro-optic and thermo-optic effects in silicon, there has been intense interest in hybrid structures in which that switching could be accomplished by integrating another material into the waveguide, including the phase-changing material, vanadium dioxide (VO2). It has long been known that the phase transition in VO2 can be triggered by ultrafast laser pulses, and that pump-laser fluence is a critical parameter governing the recovery time of thin films irradiated by femtosecond laser pulses near 800 nm. However, thin-film experiments are not a priori reliable guides to using VO2 for all-optical switching in on-chip silicon photonics because of the large changes in VO2 optical constants in the telecommunications band, the requirement of low insertion loss, and the limits on switching energy permissible in integrated photonic systems. Here we report the first measurements to show that the reversible, ultrafast photo-induced phase transition in VO2 can be harnessed to achieve sub-picosecond switching when small VO2 volumes are integrated in a silicon waveguide as a modulating element. Switching energies above threshold are of order 600 fJ/switch. These results suggest that VO2 can now be pursued as a strong candidate for all-optical switching with sub-picosecond on-off times.

physics.optics↗

arXiv2017

Silicon waveguide modulator with embedded phase change material

Phase-change materials (PCMs) have emerged as promising active elements in silicon (Si) photonic systems. In this work, we design, fabricate, and characterize a hybrid Si-PCM optical modulator. By integrating vanadium dioxide (a PCM) within a Si photonic waveguide, in a non-resonant geometry, we demonstrate ~ 10 dB broadband modulation with a PCM length of 500 nm.

physics.app-ph↗