User profiles for N. Maheswaranathan
Niru MaheswaranathanMeta Reality Labs Verified email at fb.com Cited by 5806 |
Deep unsupervised learning using nonequilibrium thermodynamics
…, E Weiss, N Maheswaranathan… - International …, 2015 - proceedings.mlr.press
A central problem in machine learning involves modeling complex data-sets using highly
flexible families of probability distributions in which learning, sampling, inference, and …
flexible families of probability distributions in which learning, sampling, inference, and …
Learned optimizers that scale and generalize
O Wichrowska, N Maheswaranathan… - International …, 2017 - proceedings.mlr.press
… are based not on the gradient evaluated at the current iterate ✓n, but rather at a location φn
which is extrapolated ahead of … At each training step n the attended location is set as φn+1 …
which is extrapolated ahead of … At each training step n the attended location is set as φn+1 …
Deep learning models of the retinal response to natural scenes
L McIntosh, N Maheswaranathan… - Advances in neural …, 2016 - proceedings.neurips.cc
A central challenge in sensory neuroscience is to understand neural computations and
circuit mechanisms that underlie the encoding of ethologically relevant, natural stimuli. In …
circuit mechanisms that underlie the encoding of ethologically relevant, natural stimuli. In …
[PDF][PDF] A multiplexed, heterogeneous, and adaptive code for navigation in medial entorhinal cortex
… For each null dataset, we randomly selected N cells as ‘‘mixed selective’’ (N = 292 cells,
the number of MS cells in the original dataset, drawn from the entire dataset of 794 cells), re-…
the number of MS cells in the original dataset, drawn from the entire dataset of 794 cells), re-…
[PDF][PDF] Social control of hypothalamus-mediated male aggression
T Yang, CF Yang, MD Chizari, N Maheswaranathan… - Neuron, 2017 - cell.com
How environmental and physiological signals interact to influence neural circuits underlying
developmentally programmed social interactions such as male territorial aggression is …
developmentally programmed social interactions such as male territorial aggression is …
Understanding and correcting pathologies in the training of learned optimizers
Deep learning has shown that learned functions can dramatically outperform hand-designed
functions on perceptual tasks. Analogously, this suggests that learned optimizers may …
functions on perceptual tasks. Analogously, this suggests that learned optimizers may …
Universality and individuality in neural dynamics across large populations of recurrent networks
N Maheswaranathan, A Williams… - Advances in neural …, 2019 - proceedings.neurips.cc
Many recent studies have employed task-based modeling with recurrent neural networks (RNNs)
to infer the computational function of different brain regions. These models are often …
to infer the computational function of different brain regions. These models are often …
Meta-learning update rules for unsupervised representation learning
A major goal of unsupervised learning is to discover data representations that are useful for
subsequent tasks, without access to supervised labels during training. Typically, this …
subsequent tasks, without access to supervised labels during training. Typically, this …
[PDF][PDF] Interpreting the retinal neural code for natural scenes: From computations to neurons
Understanding the circuit mechanisms of the visual code for natural scenes is a central goal
of sensory neuroscience. We show that a three-layer network model predicts retinal natural …
of sensory neuroscience. We show that a three-layer network model predicts retinal natural …
Guided evolutionary strategies: Augmenting random search with surrogate gradients
… In addition, we never need to work with or compute a full n × n covariance matrix. Finally,
Hansen (2011) proposes a modification to CMA-ES that accommodates external information by …
Hansen (2011) proposes a modification to CMA-ES that accommodates external information by …