Spiking Neural Network Hits 1B Parameters, Hints at New Behavior A spiking neural network allegedly reached 1.088 billion parameters and trained from random initialization to a reported loss of 4.4 before the builder ran out of money. That is the claim. The more interesting question is whether the evidence says pure SNN language models are finally leaving the toy-demo phase. I started out expecting the answer to be “not really.” Most prior work says direct training is the hard part.

A spiking neural net just hit 1.088B params—and the results may change what we think SNNs can do. Real breakthrough or hype? Click to see. #SpikingNeuralNetworks #AGI #LargeLanguageModels

Advances in memristor based artificial neuron fabrication-materials, models, and applications

Exploring the evolution of #SpikingNeuralNetworks, this review discusses #Memristor-based #ArtificialNeurons and their integration with #Sensors to realize energy-efficient, hardware-implemented brain-like intelligence.

#OpenAccess: doi.org/10.1088/2631...

„Vor wenigen Wochen stellte ein chinesisches Forscherteam das Modell „SpikingBrain 1.0“ vor – eine #KI auf Basis eines #SpikingNeuralNetworks. Diese Technik soll nicht nur weniger Energie verbrauchen, sondern auch ohne Nvidia-Chips und ohne große Datenmengen auskommen.“🤔
Lina Knees via #Handelsblatt

Self‑Distillation Boosts Efficient Spiking Neural Network Training

A self‑distillation framework speeds SNN training and cuts memory use, achieving competitive accuracy on CIFAR‑10, CIFAR‑100 and ImageNet; code on GitHub. Read more: getnews.me/self-distillation-boosts... #spikingneuralnetworks #selfdistillation

Random Feature Spiking Neural Networks Offer New Training Path

Study introduces S-SWIM, a random-feature training for spiking neural networks that needs only one data pass and matches state-of-the-art accuracy; submitted 1 Oct 2025. getnews.me/random-feature-spiking-n... #sswim #spikingneuralnetworks

Spiking Neural Networks with Attention Boost Remote Sensing Super-Resolution

SpikeSR with a Spiking Attention Block beat prior methods on AID, DOTA and DIOR benchmarks; the paper was accepted at a 2025 ML conference. Read more: getnews.me/spiking-neural-networks-... #spikingneuralnetworks #attention #remotesensing

DelRec Enables Learning of Delays in Recurrent Spiking Neural Networks

DelRec is a method that learns weights and axonal delays in spiking neural networks, reaching high accuracy on Speech Commands and Permuted Sequential MNIST. Code on GitHub. getnews.me/delrec-enables-learning-... #spikingneuralnetworks #delrec

Hybrid ANN‑SNN Framework Uses Surrogate Spike Encoding

Scientists introduced a hybrid ANN‑SNN model with a surrogate gradient for bit‑plane spike encoding, allowing end‑to‑end training. The pre‑print appeared in September 2025. Read more: getnews.me/hybrid-ann-snn-framework... #hybridannsn #spikingneuralnetworks

Spiking Neural Networks Naturally Sparse Gradients Enhance Robustness

Researchers find spiking neural network designs produce sparse gradients, giving robustness without regularization, reducing generalization on clean data. Read more: getnews.me/spiking-neural-networks-... #spikingneuralnetworks #sparsity

Spiking Neural Networks Gain Universal Approximation Theory

Researchers proved that spiking neural networks can universally approximate any continuous function on compact domains, and the pre‑print was submitted on 26 Sep 2025. getnews.me/spiking-neural-networks-... #spikingneuralnetworks #neuromorphic

SpikeMatch: Semi‑Supervised Learning for Spiking Neural Networks

SpikeMatch, a technique for spiking neural networks, creates pseudo‑labels from weakly‑augmented data and improves results, keeping energy use low. Read more: getnews.me/spikematch-semi-supervis... #spikematch #spikingneuralnetworks

Spiking Neural Networks Enable Low-Power Mental Workload Detection

Spiking neural networks classify mental workload with accuracy to traditional models while using less power, thanks to event‑driven processing and multimodal sensor data. getnews.me/spiking-neural-networks-... #spikingneuralnetworks #mentalworkload

New Refractory-Period Spiking Neural Network Boosts AI Performance

RPLIF raises a neuron's firing threshold after each spike, adding a refractory window. It records 82.40% accuracy on CIFAR‑10‑DVS and 97.22% on DVS128 Gesture. Read more: getnews.me/new-refractory-period-sp... #spikingneuralnetworks #rplif

SPACE Method Boosts Test‑Time Adaptation for Spiking Neural Networks

SPACE is a source‑free test‑time adaptation for spiking neural networks that aligns feature maps across augmented views of a test sample, keeping low‑power operation. getnews.me/space-method-boosts-test... #spikingneuralnetworks #testtimeadaptation

Predictive Spike Timing Enables Shortest Path Computation in Networks

A new predictive spike‑timing algorithm enables spiking neural networks to compute shortest paths using only local neuron connections; the study was submitted on 12 September 2025. getnews.me/predictive-spike-timing-... #spikingneuralnetworks

Edge Intelligence with Spiking Neural Networks
Albert Y. Zomaya, Changze Lv et al.
Paper
Details
#EdgeAI #SpikingNeuralNetworks #ZomayaResearch

#Neuromonster #WolfgangMaass #NeuroAI #SpikingNeuralNetworks #AIandBrains #KeynoteSpeaker #NeuroscienceMeetsAI

Robot deployment feasibility study: Left: Real-time deployment of our Modular SNN in a small indoor environment on a CPU. First, the reference dataset is collected and the Modular SNN is trained offline. During inference, the robot moves through the environment, collecting images, and predicting their place labels in real time. The reference and query images are collected at different times of the day. The images were collected at 1 Hz with the robot moving at approximately 1 m/s. The blue line and points represent the reference path and images, while the green line represents the query path. Correct predictions are marked with green crosses, and incorrect predictions are marked with red crosses. The four samples of the query and predicted places show the original images and their preprocessed forms, which are used as input to our Modular SNN. Right: Proof-of-concept robot deployment testing platform. We used an AgileX Scout Mini [60] robot equipped with an Intel RealSense D435 camera.

Researchers from Queensland University of Technology present an energy-efficient #placerecognition system leveraging Spiking Neural Networks with modularity and sequence matching to rival traditional deep networks
ieeexplore.ieee.org/document/107...

#SpikingNeuralNetworks #RobotPerception

Sydney Harbour Bridge. CC BY NC 2024 Dylan Muir

Hi! 👋 I am working to make #SpikingNeuralNetworks the next big thing for #MachineLearning. Currently I'm at SynSense, focussed on applications for #SNNs, as well as toolchains to enable ML and SW Engineers to use our #Neuromorphic technology for low-power sensing and processing.

SNNs In Vision: Event-Based Processing For Real-Time AI SNNs power real-time perception in vision systems by enabling event-based processing, offering faster, more efficient AI for dynamic environments.

Spiking Neural Networks (SNNs) are creating a buzz in the world of artificial intelligence and neuromorphic engineering. #eventbasedprocessing #eventdrivenAI #Perception #realtimeAI #realtimevision #SNNs #spikingneuralnetworks #visionsystems
aicompetence.org/snns-in-visi...