The field of neural networks and control systems is rapidly evolving, with a focus on developing more efficient, stable, and adaptive systems. Recent research has explored the use of spiking neural networks, which mimic the behavior of biological neurons, for applications such as continuous control and regression tasks. Additionally, there has been a push towards developing more robust and generalizable neural network architectures, including those that incorporate variance suppression and sharpness-aware minimization. In the realm of control systems, researchers have been investigating the use of bio-inspired approaches, such as spiking control systems and neuromorphic controllers, for applications such as soft robotics and industrial automation. Noteworthy papers in this area include the introduction of optimized weight initialization on the Stiefel manifold for deep ReLU neural networks, which has been shown to improve training stability and generalization performance. Another notable work is the development of a brain-inspired gating mechanism for spiking neural networks, which has been demonstrated to enhance robustness and computational efficiency.
Advancements in Neural Networks and Control Systems
Sources
Optimized Weight Initialization on the Stiefel Manifold for Deep ReLU Neural Networks
Design, Modelling and Analysis of a Bio-inspired Spiking Temperature Regulator
End-to-End Low-Level Neural Control of an Industrial-Grade 6D Magnetic Levitation System
VISP: Volatility Informed Stochastic Projection for Adaptive Regularization
Guidance and Control Neural Network Acceleration using Memristors
Nuclear fusion plasma fuelling with ice pellets using a neuromorphic controller
VASSO: Variance Suppression for Sharpness-Aware Minimization
The Lifecycle Principle: Stabilizing Dynamic Neural Networks with State Memory
Spiking control systems for soft robotics: a rhythmic case study in a soft robotic crawler
LSAM: Asynchronous Distributed Training with Landscape-Smoothed Sharpness-Aware Minimization
Vibration Damping in Underactuated Cable-suspended Artwork -- Flying Belt Motion Control
A Brain-Inspired Gating Mechanism Unlocks Robust Computation in Spiking Neural Networks
Initialization Schemes for Kolmogorov-Arnold Networks: An Empirical Study
Insights from Gradient Dynamics: Gradient Autoscaled Normalization
Spiking Neural Network Decoders of Finger Forces from High-Density Intramuscular Microelectrode Arrays
Memristor-Based Neural Network Accelerators for Space Applications: Enhancing Performance with Temporal Averaging and SIRENs
Depth-Aware Initialization for Stable and Efficient Neural Network Training
Spiking Neural Networks for Continuous Control via End-to-End Model-Based Learning
STL-based Optimization of Biomolecular Neural Networks for Regression and Control
SuperSNN: A Hardware-Aware Framework for Physically Realizable, High-Performance Superconducting Spiking Neural Network Chips
Genesis: A Spiking Neuromorphic Accelerator With On-chip Continual Learning
Hardware Acceleration of Kolmogorov-Arnold Network (KAN) in Large-Scale Systems
Full Integer Arithmetic Online Training for Spiking Neural Networks