Neural Networks and Reservoir Computing

Information coding strategies in biological reservoir computing for multiclass time-series classification

The paper investigates how input encoding and spike detection methods affect neural representations and reservoir computing performance in biological neural networks for multiclass time-series classification.

CLEAR-FORCE: A reservoir computing framework for closed-loop brain stimulation

The paper introduces CLEAR-FORCE, a reservoir computing framework for real-time closed-loop brain stimulation, showing success in inducing and suppressing neural oscillations through simulations.

Plasticity Dynamics Improves Reservoir Performance on Mixed-Timescale Memory Tasks

This paper introduces synaptic plasticity into recurrent networks to improve their ability to process mixed-timescale signals by enabling simultaneous handling of fast and slow components.

Learning to play with spikes: characterizing, predicting, and engineering unsupervised plasticity rules for spiking reservoir computing

This paper characterizes and predicts effective unsupervised plasticity rules for spiking reservoir computing by analyzing their performance on a Pong game task.

Brain organoid computing for robotic decision-making

This paper introduces Brainobot, a biohybrid robotic system using brain organoid reservoir computing to enable real-world decision-making tasks like object grasping and laser chasing with high efficiency and low energy consumption.

Harnessing the Intrinsic Dynamics of Biological Neural Networks for Reservoir Computing

This paper demonstrates that neuronal cultures can function as living reservoirs for reservoir computing, achieving high XOR accuracy and retaining state information despite neural drift.

Nonlinear regime transitions enable reservoir computation in activator-inhibitor cellular automata

This paper explores how activator-inhibitor cellular automata can function as reservoirs for computation by tuning nonlinear dynamics to enable memory and separability.

Decision Confidence Neuron in Echo State Network for Continual Evaluation of EEG Motor Imagery Classification Quality

The paper introduces a decision confidence neuron for ESNs to continuously monitor classification confidence in EEG motor imagery tasks, showing it correlates with decision accuracy and discriminates correct from error decisions.

Handwritten Digit classification with neural cultures is influenced by neural architecture, network dynamics, and decoding methods

This study evaluates how neural architecture, network dynamics, and decoding methods affect handwritten digit classification using human iPSC-derived neuronal networks.

Modeling the rhythmic complexity of professional drumming with an oscillation-driven reservoir computer

This paper uses an oscillation-driven reservoir computer to model professional drumming rhythms, showing it can replicate timing and amplitude patterns of human drumming.

Living Bacterial Reservoir Computers for Information Processing and Sensing

This paper demonstrates using E. coli as a living reservoir computer to classify COVID-19 plasma samples and perform nonlinear computations through growth responses.

Neuromorphic hierarchical modular reservoirs

This paper introduces hierarchical modular networks as reservoirs to study computational advantages of brain-like modularity, showing enhanced memory, multitasking, and temporal dynamics.

Morphologically Tunable Mycelium Chips for Physical Reservoir Computing

The paper introduces biodegradable mycelium-based chips for physical reservoir computing, demonstrating their ability to perform machine learning tasks with sustainability and cost advantages.

Theory of Temporal Pattern Learning in Echo State Networks

The paper develops a quantitative theory to explain learning in echo state networks with stable dynamics and weak feedback, showing that learning can be described by nonlinear interactions of master modes.

Non-random brain connectome wiring enables robust and efficient neural network function under high sparsity

The paper shows that non-random brain connectome wiring improves robustness and efficiency in sparse neural networks compared to random wiring.

Linearizing and forecasting: a reservoir computing route to digital twins of the brain

This paper uses reservoir computing with noisy linear recurrent neural networks to create digital twins of brain activity from resting-state fMRI data, enabling prediction and decomposition of BOLD signals into low-dimensional latent states.

A novel framework for expanding RNNs with biophysical detail to solve cognitive tasks

This paper introduces a biophysically detailed reservoir computing framework to study how biophysical properties affect cognitive task performance in neural networks.

Controlling Integration and Segregation in Echo State Networks via Noradrenaline and Acetylcholine Neuromodulation

This paper introduces a modular echo state network with neuromodulation to control integration and segregation, showing improved performance on context-dependent tasks.

Reservoir Computing with Ultra-Sparse Rings

This paper shows that ultra-sparse ring-connected reservoirs can perform complex dynamical systems tasks with linearly scaling parameters.

Topological Entropy Correlates with the Predictive Power of Multiplexed Ensemble Reservoir Computing

This paper introduces DynML, a dynamical systems framework for modeling biological time series, showing that reservoir topological entropy correlates with predictive power.

Cross-species connectome comparisons reveal the network attributes of memory capacity and time series prediction

The paper uses connectomes from four species in echo state networks to show that brain-like architectures improve time series prediction and memory capacity compared to random networks.

A Mean-Field Approach to Criticality in Spiking Neural Networks for Reservoir Computing

This paper develops a mean-field framework to configure spiking neural networks in the critical regime for reservoir computing, showing improved performance and computational capacity near criticality.

Decreased spinal inhibition leads to undiversified locomotor patterns

This study investigates how the balance of excitation and inhibition in neural networks affects the learning of human locomotor patterns using reservoir computing.

Selective consistency of recurrent neural networks induced by plasticity as a mechanism of unsupervised perceptual learning

This paper shows that a plastic neural network can acquire selective consistency to repeated stimuli through self-organization, even without performance improvement.

conn2res: A toolbox for connectome-based reservoir computing

conn2res is a Python toolbox that translates connectomes into artificial neural networks to study brain network computation.

TIME-SERIES CLASSIFICATION FOR PATIENTS UNDER ACTIVE SURVEILLANCE AND SCREENING USING ECHO STATE NETWORK

This paper develops an echo state network methodology to classify time-series data from patients under active surveillance and screening, addressing heterogeneity in visit intervals.

Brain Connectivity meets Reservoir Computing

This study explores how real brain connectivity patterns can inform artificial neural network architectures, finding that biologically inspired networks perform as well as classical ones with sufficient randomness and diversity in connections.

Effects of Connectivity on Narrative Temporal Processing in Structured Reservoir Computing

The paper uses structured reservoir computing to model how white matter connectivity influences narrative processing speed in different cortical areas.

Baseline control of optimal performance in recurrent neural networks

This study investigates how baseline modulations in recurrent neural networks control dynamical phases and achieve optimal performance without fine-tuning recurrent couplings.

Flexible cognition in context-modulated reservoir networks

This paper shows that rapid multitask learning in reservoir-style networks can occur with sparse, localized plasticity and local error signals, without backpropagation.