Tolmachev Pavel (@TolmachevPavel2) header
Tolmachev Pavel (@TolmachevPavel2) profile picture

Tolmachev Pavel

@TolmachevPavel2

Researcher in Computational Neuroscience @PrincetonNeuro with focus on neural representations, recurrent neural networks and reinforcement learning

 Princeton
Joined April 2016
  • Tweets5
  • Following131
  • Followers127
Excited to share our work with @EngelTatiana now published in Nature Machine Intelligence! nature.com/articles/s42256-0…
Image from Tolmachev Pavel (@TolmachevPavel2)
Tolmachev Pavel (@TolmachevPavel2) profile pictureTolmachev Pavel@TolmachevPavel226 Nov 2024
RNNs are often used as a model for exploring how the brain may solve specific tasks. In a new preprint, we show that, depending on the architecture, RNNs find different circuit solutions, behaving differently when exposed to novel stimuli. biorxiv.org/content/10.1101/… @EngelTatiana
Single-unit activations confer inductive biases for emergent circuit solutions to cognitive tasks

Single-unit activations confer inductive biases for emergent circuit solutions to cognitive tasks

Trained recurrent neural networks (RNNs) have become the leading framework for modeling neural dynamics in the brain, owing to their capacity to mimic how population-level computations arise from...

biorxiv.org
2
20
2
94
14,444
RNNs are often used as a model for exploring how the brain may solve specific tasks. In a new preprint, we show that, depending on the architecture, RNNs find different circuit solutions, behaving differently when exposed to novel stimuli. biorxiv.org/content/10.1101/… @EngelTatiana
Single-unit activations confer inductive biases for emergent circuit solutions to cognitive tasks

Single-unit activations confer inductive biases for emergent circuit solutions to cognitive tasks

Trained recurrent neural networks (RNNs) have become the leading framework for modeling neural dynamics in the brain, owing to their capacity to mimic how population-level computations arise from...

biorxiv.org
1
18
3
108
29,798
The choice of activation function in RNNs is often assumed to minimally affect its trajectories. We analyzed ReLU, sigmoid, and tanh RNNs on diverse tasks, revealing differences in their neural trajectories and individual neuron responses, challenging this assumption.
Image from Tolmachev Pavel (@TolmachevPavel2)
2
2
1
9
3,085
Load more