News
*PhD and Postdoc Job Alert*
The positions will focus on developing an accelerator for the simulation of cortical models of processing and learning.
Review of applications will begin immediately and continue until the position is filled.
2025
Highlight from the Forum für Universität und Gesellschaft - Info
The Forum provides free public access to expert knowledge to compile, clarify, and evaluate current information on selected topics. The 2024/25 series of event focuses on human insights in the era of artificial intelligence.
Our Mihai A. Petrovici explained how insights into brain learning mechanisms inspire the development of more efficient AI systems.
His presentation is a highlight of the Forum.
A brochure of the whole event is now available here.
For references to our contributions, see especially pages 3, 11, and 18-19.
Capo Caccia Cognitive Neuromorphic Engineering Workshop - April 27 - May 11 - Info
> Mihai A. Petrovici gave a lecture on "Computing with Spikes"
> Paul Haider and Timo Gierlich co-presented "Learning to control a cartpole with NN-based controllers"
As always, the science was great and it was a lot of fun!
New contribution to a Preprint!
"Neural mechanisms of predictive processing: a collaborative community experiment through the OpenScope program"
-> Publication
New contribution to a Preprint!
"Building functional and mechanistic models of cortical computation based on canonical cell type connectivity"
-> Publication
Design, Automation and Test in Europe Conference - Mar. 31 - Apr.2 - Info
COSYNE Conference & Workshop - Mar. 27 - Apr. 1 - Info
Two group members presented a Poster!
- Prospective and retrospective coding in cortical neurons by Simon Brandt
- Backpropagation through space, time and the brain by Paul Haider
NICE Conference - Mar 25-28 - Info
Four flash talks were presented!
- Weight transport through spike timing for robust local gradients by Timo Gierlich
see presentation - Backpropagation through space, time and the brain by Benjamin Ellenberger
- DelGrad: Exact gradients in spiking networks for learning transmission delays and weights by Jimmy Weber
- Elise: efficient learning of sequences in structured recurrent networks by Ben von Hünerbein - see presentation
Advancing AI with Brain-Inspired Computing: new contribution in ScienceNews!
Mihai A. Petrovici discussed the crucial role of neuromorphic hardware in the future of artificial intelligence and in enhancing AI efficiency and adaptability.
Read the article: More brainlike computers could change AI for the better
New Preprint!
"Weight transport through spike timing for robust local gradients"
-> Publication
New NSF-SNSF Lead agency funded Project! - Start: 01.04.2025 / Runtime: 3 years
The NSF-SNSF Lead Agency process allows researchers in Switzerland and the US to submit a joint research proposal to either national funding agency.
The MINT project entitled "NSF-SNSF: Scalable accelerated learning in continuous neuronal systems" was awarded a total of $1,178,194.
$400,000 was awarded to the US collaborator Prof. Rajit Manohar and
CHF 682,624 to the project leader Dr Mihai A. Petrovici.
New Article contribution in Nature Communications!
Bringing together members of academia and industry to define benchmarks for "brain-inspired" neuromorphic computing.
"The neurobench framework for benchmarking neuromorphic computing algorithms and systems"
-> Publication and Website
Swiss Computational Neuroscience Retreat - Feb. 5-7 - Info
Three talks were presented!
> Backpropagation through space, time and the brain by Paul Haider
> Prospective and retrospective coding in cortical neurons by Simon Brandt
> Elise: Efficient learning of Sequences in Structured Recurrent Networks
by Ben von Hünerbein
New Article in Trends in Neuroscience and Education!
Introducing Printed Circuit Board that mimics the basic dynamics of neurons.
"Lu.i – A low-cost electronic neuron for education and outreach" - For more info, -> Lu.i page and the Publication
New Preprint!
From milliseconds to nanoseconds: photonic spiking nanolasers enable a dramatic acceleration of Bayesian inference via sampling.
"Ultrafast neural sampling with spiking nanolasers"
-> Publication
Lectured by Lu.i!
Undergraduate students at the Technische Hochschule Nürnberg Georg Simon Ohm developed academic projects using our Lu.i neurons, which Prof. Axenie later showcased in his lecture on neuromorphic systems!
For more info, see the academic neuromorphic projects:
Hybrid ArtificialNN and SpikingNN for closed-loop motor control and Spiking Silicon Neurons for Text2Morse Conversion
2024
New contribution in eLife!
Searching the most efficient path to real-time learning
"A Neuronal Least-action Principle For Real-time Learning In Cortical Circuits"
-> Publication- Digest
11th Christmas Symposium & Party - Department of Physiology - Info
Arno Granier with "Building models of the cortex from empirical gifts: theoreticians and experimentalists celebrating together"
and Julian Göltz with "Twinkle twinkle little neuron"
delivered Christmas-themed presentations to spread the festive spirit!
Spiking Neural networks as Universal Function Approximators Workshop (SNUFA) - Nov. 5-6 - Info
>Timo Gierlichpresented the Flash Talk Weight transport through spike timing for robust local gradients
> Maryeme Ouafoudipresented the Poster Evolving Interpretable Plasticity Rules on Accelerated Neuromorphic Hardware
> Julian Göltzand Jimmy Weber presented the Poster DelGrad: Exact gradients in spiking networks for learning transmission delays and weights
Forum für Universität und Gesellschaft - Nov. 02 - Info
The Forum provides free public access to expert knowledge to compile, clarify, and evaluate current information on selected topics. The 2024/25 series of event focuses on human insights in the era of artificial intelligence.
Watch Mihai A. Petrovici explaining how insights into brain learning mechanisms inspire the development of more efficient AI systems.
Das Gehirn als Vorbild: künstliche neuronale Netzwerke.
Interview by the Schweizer Radio und Fernsehen (SRF) Wissenschaftsmagazin - Info
Mihai A. Petrovici said a few words about the 2024 Nobel Prize in Physics (9:57 - 10:18)
Bernstein Conference & Workshops 2024 - Sep. 29- Oct. 2 - Info
> Mihai A. Petrovici presented the talks "Error propagation through space, time and the brain" and "Computation and learning in physical neuronal systems"
> Kevin Max presented the talk "Biological deep learning with three-component rules"
> Katharina Wilmes presented the talk "Uncertainty-modulation of plasticity"
> Peter Lake presented the poster "DelGrad: Exact gradients in spiking networks for learning transmission delays and weights"
> Julian Göltz presented the poster "Emergence of Synfire Chains in Functional Multi-Layer Spiking Neural Networks"
New Article contribution in PNAS Nexus!
"Confidence and second-order errors in cortical circuits"
-> Publication
AREADNE 2024 Research in Encoding And Decoding of Neural Ensembles - June 25-29 - Info
Federico Benitez presented "Order from chaos: Interplay of development and learning in recurrent networks of structured neurons" (Poster)
Manfred-Stärk-Symposium 2024 - Thunersee - June. 11-13
Special symposium on Theoretical Neuroscience and Neuromorphic Computation dedicated to our generous benefactor Manfred Stärk and his foundation.
New Article! Single neurons perform bayes-optimal cue integration.
"Conductance-based dendrites perform Bayes-optimal cue integration"
-> Publication
New Article: (PAL), a bio-plausible method to learn efficient feedback weights
"Learning efficient backprojections across cortical hierarchies in real time"
-> Publication
ICNCE: International Conference on Neuromorphic Computing and Engineering - June 3-6 - Info
> Katharina Wilmes presented the Poster "Confidence estimation and second-order errors in cortical circuits" and the Talk "Neural mechanisms underlying predictive processing in uncertainty"
> Timo Gierlich presented the Poster "Weight transport through spike timing for robust local gradients"
> Peter Lake presented the Poster "DelGrad: Training transmission delays for spiking networks on neuromorphic hardware"
> Paul Haider gave the Talk "Backpropagation through space, time and the brain"
New Article Contribution!
"How Adversarial REM Dreams May Facilitate Creativity, and Why We Become Aware of Them"
-> Publication
Neuromonster: Fifth International Convention on the Mathematics of Neuroscience
and AI - May 28 - 31 - Info
> Arno Granier gave the Talk "Confidence estimation and second-order errors in cortical circuits" -> recording
> Paul Haider gave the spotlight Talk "Backpropagation through space, time and the brain" -> recording
> Simon Brandt presented the Poster "Prospective and Retrospective Coding in Cortical and Hippocampal Pyramidal Neurons"
ZEN Neuroscience Lecture - May 24
Mihai A. Petrovici gave the lecture "Computing with physics: Intelligence in carbon and silicon" - Abstract
New Preprint Contribution!
"Prospective and retrospective coding in cortical neurons"
-> Publication
EMBO Workshop Principles of dendritic function and computation - May 21-24 - Info
Ben von Hünerbein presented "Order from chaos: Interplay of development and learning in recurrent networks of structured neurons" (Flash Poster Talk)
LAKE CONFERENCES: The Neurobiology of Mental Health - May 12-16 - Info
Katharina Wilmes presented "Uncertainty-weighted prediction errors (UPEs) in cortical microcircuits" (Poster)
New Preprint! An efficient gradient-based algorithm for training temporal variables, specifically transmission delays in SNNs alongside synaptic weights.
"DelGrad: Exact gradients in spiking networks for learning transmission delays and weights" -
-> Publication
Capo Caccia Cognitive Neuromorphic Engineering Workshop - April 28 - May 12 - Info
> Mihai A. Petrovici gave a lecture on "Why neurons spike: SNN applications"
> Julian Göltz co-presented "NeuroMorsick, interactive Morse decoding with DelGrad"
> Ismael Jaras led a detailed discussion group on "Fully local learning through space and time" and participated together with Maryeme Ouafoudi to the workgroup "Learning for Neuromorphic Hardware"
> Mihai and Ismael also encouraged the practice of adversarial training
As always, the science was great and it was a lot of fun!
New Preprint! Introducing Printed Circuit Board mimicking the basic dynamics of neurons
"Lu.i - A low-cost electronic neuron for education and outreach"
-> Publication- Lu.i page
New Contribution to a multi-disciplinary review on the ethics of brain-inspired AI
"A method for the ethical analysis of brain-inspired AI"
-> Publication
PLENA - Pucón Learning and AI Summit - Apr. 8-12 - Info
New Preprint! Generalized Latent Equilibrium (GLE) computational framework for fully local spatio-temporal credit assignment in physical, dynamical networks of neurons.
"Backpropagation through space, time, and the brain"
-> Publication
New Preprint!
"Increased perceptual reliability reduces membrane potential variability in cortical neurons"
-> Publication
New Preprint!
"Order from chaos: Interplay of development and learning in recurrent networks of structured neurons"
-> Publication
AIroV Symposium on AI, Robotics, and Vision - Mar. 26-27 - Info
COSYNE Conference & Workshop - Feb. 29 - Mar 5 - Info
Five group members presented a Poster!
- Ben von Hünerbein:"Order from chaos: Interplay of development and learning in recurrent networks of structured neurons"
- Laura Kriener:"Learning efficient backprojections across cortical hierarchies in real time"
- Timo Gierlich:"Learning efficient backprojections across cortical hierarchies in real time"
- Julian Göltz:"Gradient-based methods for spiking physical systems"
and Katharina Wilmesco-organised the workshop "The predictive brain: from algorithms to mechanisms"
CAIM Ethics Talks - Feb. 29 - Info
The CAIM Embedded Ethics Lab hosted its second series of Ethics Talks entitled "Transparency and Ethics in AI Development" to foster public dialogue into ethical challenges of AI applications in medicine.
Mihai A. Petroviciwas invited to discuss "AI, autonomy, and mental health." -> Recording
Swiss Computational Neuroscience Retreat - Jan 31 - Feb. 2
Ismael Jaras presented "Learning, forward and backward" and Arno Granier "Confidence estimation and second-order errors in cortical circuits".
EBRAINS 2.0 - New funded Project! - Start: 01.01.2024 / Runtime: 3 years
The EBRAINS 2.0 project was submitted in response to the INFRASERV call (a Horizon Europe's Research Infrastructures actions) and was awarded €38 million to further develop the EBRAINS research infrastructure services, a legacy of the Human Brain Project.
The project involves 59 partner institutions from 16 European countries and officially started in January 2024. As partners, our group together with CompNeuro were granted €1.1 million.