#rnns — Public Fediverse posts

🧠⭕️ New preprint by Hulse et al: How can the #fly’s neural compass show #RingAttractor dynamics despite heterogeneous #connectomic wiring? They combine task-trained #RNNs, theory and fly #connectomes to show that “dynamical clones” can embed hidden symmetries in apparently messy connectivity, preserving stable head-direction #attractors and angular velocity integration.

🌍 https://doi.org/10.64898/2026.05.18.725766

#Neuroscience #CompNeuro #NeuralDynamics #NeuralManifolds #RingAttractor

🧠 New paper by Deistler et al: #JAXLEY: differentiable #simulation for large-scale training of detailed #biophysical #models of #NeuralDynamics.

They present a #differentiable #GPU accelerated #simulator that trains #morphologically detailed biophysical #neuron models with #GradientDescent. JAXLEY fits intracellular #voltage and #calcium data, scales to 1000s of compartments, trains biophys. #RNNs on #WorkingMemory tasks & even solves #MNIST.

🌍 https://doi.org/10.1038/s41592-025-02895-w

#Neuroscience #CompNeuro

🧠 New preprint by Codol et al. (2025): Brain-like #NeuralDynamics for #behavioral control develop through #ReinforcementLearning. They show that only #RL, not #SupervisedLearning, yields neural activity geometries & dynamics matching monkey #MotorCortex recordings. RL-trained #RNNs operate at the edge of #chaos, reproduce adaptive reorganization under #visuomotor rotation, and require realistic limb #biomechanics to achieve brain-like control.

🌍 https://doi.org/10.1101/2024.10.04.616712

#CompNeuro #Neuroscience

#ITByte: The #MachineLearning models having sequential data as input or output are called #SequenceModels.

It includes text streams, video clips, audio clips, time-series data, etc. Recurrent Neural Networks (#RNNs) and Long Short-Term Memory(#LSTM) are popular algorithms used in sequence models.

https://knowledgezone.co.in/trends/explorer?topic=Sequence-Model

#ITByte: The #MachineLearning models having sequential data as input or output are called #SequenceModels.

It includes text streams, video clips, audio clips, time-series data, etc. Recurrent Neural Networks (#RNNs) and Long Short-Term Memory(#LSTM) are popular algorithms used in sequence models.

https://knowledgezone.co.in/trends/explorer?topic=Sequence-Model

Learning better with Dale’s Law: A Spectral Perspective - #NeurIPS2023 contribution by Li et al. (2023). It discusses how to train brainlike #RNNs with separate inhibitory and excitatory units with similar performance as standard RNNs:

🌍 https://openreview.net/forum?id=rDiMgZulwi

#RNN #DalesLaw #CompNeuro #Neuroscience

1997 with the advent of Long Short-Term Memory recurrent #neuralnetworks marks the subsequent step in our brief history of )large) #languagemodels from last week's #ise2023 lecture. Introduced by Sepp Hochreiter and Jürgen Schmidhuber #LSTM #RNNs enabled efficient processing of sequences of data.
Slides: https://drive.google.com/file/d/1atNvMYNkeKDwXP3olHXzloa09S5pzjXb/view?usp=drive_link
#nlp #llm #llms #ai #artificialintelligence #lecture @fizise

#ITByte: The #MachineLearning models having sequential data as input or output are called #SequenceModels.

It includes text streams, video clips, audio clips, time-series data, etc. Recurrent Neural Networks (#RNNs) and Long Short-Term Memory(#LSTM) are popular algorithms used in sequence models.

https://knowledgezone.co.in/trends/explorer?topic=Sequence-Model

'Minimal Width for Universal Property of Deep RNN', by Chang hoon Song, Geonho Hwang, Jun ho Lee, Myungjoo Kang.

http://jmlr.org/papers/v24/22-1191.html

#rnns #rnn #deep

Investigating Action Encodings in Recurrent Neural Networks in Reinforcement Learning

Matthew Kyle Schlegel, Volodymyr Tkachuk, Adam M White, Martha White

https://openreview.net/forum?id=K6g4MbAC1r

#rnns #rnn #reinforcement