News Archive (2023 - 2022)

New Preprint! Path integration used by Bees for their navigation was implemented in the neuromorphic mixed-signal processor BrainScaleS-2

"Emulating insect brains for neuromorphic navigation" - See Publication

New Review in National Science Review!
The modular-integrative modelling approach shows promise in creating hybrid brain models that combine biological plausibility with functional performance.

"Modular-integrative modeling: a new framework for building brain models that blend biological realism and functional performance" - See Publication

10th Christmas Symposium & Party 2023 - Department of Physiology - Info

Ben von Hünerbein with "Jingles bells from Chaos" and
Arno Granier with "To believe or not to believe: uncertainty estimation in cortical circuits"
presented Christmas themed talks to keep up with the spirit of the symposium!

New lab member (December)

Welcome to our new PhD student Maryeme Ouafoudi!

Conclusion of the Human Brain Project (HBP)! - Nov. 21-24
Walter Senn (CompNeuro group) presented the work achieved together with the NeuroTMA group (M.A. Petrovici) during the last phase of the HBP. The SGA3 phase officially ended on 30 September. Together with our partners, we successfully completed the review of the HBP by the European Commission.
-> See Deliverable (D3.11) for an overview of our contributions to the HBP project.

SYNCH - New funded Project! - Start: 1.11.2023 / Runtime: 4 years
The collaborative SYNCH project - Combining SYnthetic Biology & Neuromorphic Computing for Chemosensory perception - has been awarded funding from the Volkswagen Foundation as part of the NEXT - Neuromorphic Computing call.
The project partners are our postdoc Kevin Max , together with Jan Steinkühler (CAU Kiel, DE) and Federico Corradi (TU Eindhoven, NL).

SYNCH will co-design synthetic biological pathways, neuromorphic electronic circuits, and bio-constrained theories of olfactory sensing to create an accurate and efficient perception system for olfaction.
The aim is to realize real-time decoding of biological interactions generated by a range of small molecules through temporal spiking data for fast, distributed, and online chemo-sensing.

International Conference on Neuromorphic, Natural and Physical Computing - Oct. 25-27
- Info

- Mihai A. Petrovici chaired the Session 3 "Guides from nature: neuroscience, theoretical biology, complex systems"
- Walter Senn gave the keynote "A theoretical physics guide to neuroscience and cognition"
- Kevin Max gave the talk "Learning efficient backprojections cross cortical hierarchies in real time"
- Julian Göltz presented the poster "Gradient-based methods for spiking physical systems"

Human intelligence meets artificial intelligence workshop - Oct. 4-5 - Info

Mihai A. Petrovici was invited to present the talk "Computing with physics: Biological and bio-inspired intelligence"

Bernstein Conference 2023 - Sep. 26-29 - Info

- Katharina A. Wilmes gave the talk "Uncertainty-modulated prediction errors in cortical microcircuits"
- Simon Brandt presented the poster "Prospective and Retrospective Coding in Cortical and Hippocampal Pyramidal Neurons"
- Arno Granier presented the poster "Reliability estimation and second-order prediction errors in cortical circuits"

New Contribution to a Preprint!

"Precision estimation and second-order prediction errors in cortical circuits" - See Publication

ICANN: Artificial Neural Networks and Machine Learning - Sep. 26-29 - Info

Kevin Max presented the talk "Learning Efficient Backprojections Across Cortical Hierarchies in Real Time"

Einstein Center for Neurosciences - NeuroCure Cluster of Excellence Student-run speaker series "Spikes - bug or feature" - Sep. 21
Mihai A. Petrovici was invited to present the lecture "Deep learning with spikes."

New lab member (September)

Welcome to Dr. Ismael Jaras, who is again joining the lab as a guest postdoc!

Jülich-Aachen Neuromorphic Computing Day - Aug.30 - Info

Mihai A. Petrovici gives his "New insights into the physics of computing"

New Contribution to a Preprint: normative approach postulating that uncertainty-weighted prediction errors are represented by layer 2/3 pyramidal neurons!

"Uncertainty-modulated prediction errors in cortical microcircuits"
- See Publication

CNS*2023: Computational Neuroscience Meeting - July. 15-19 - Info

- Katharina A. Wilmes gave the talk "Uncertainty-modulated prediction errors in cortical microcircuits"
- Mihai A. Petrovici gave the talk "Credit assignment in cortical structures"
- Kevin Max presented the poster "Learning efficient backprojections across cortical hierarchies in real time"
- Ben von Hünerbein presented the poster "Certainty-weighted integration of information in individual cortical neurons"

Lab Retreat in Langnau im Emmental - July 5-7

This year we invited our friends from the Theoretical Neuroscience group and it was another three days of science, brain-storming, sports, cooking and other team building exercises. As always, it was a lot of FUN!!

Manfred-Stärk-Symposium 2023 - Weggis - July. 2-4

Special symposium on Theoretical Neuroscience and Neuromorphic Computation dedicated to our generous benefactor Manfred Stärk and his foundation.

Cortical Computation 60th Birthday Colloquium in Honor of Prof. Walter Senn - June 15

Mihai A. Petrovici presented "Physics-inspired approaches to computational neuroscience"

Research Seminar at the Albert Einstein Center for Fundamental Physics - Bern - May 31
Mihai A. Petrovici was invited to give the talk "Computing with physics: Biological and bio-inspired intelligence."

Mihai became an Editorial Board Member of the Journal Research Directions: Bioelectronics - Info
Mihai A. Petrovici and Samit Chakrabarty wrote the following call:

"Is neuromorphic computing disruptive enough to 1) advance our understanding of the brain and 2) make the design and working of (bio)electronic devices efficient and scalable?" - See Publication

New Contribution to a Preprint: examine conceptual, technical, and ethical issues raised by the development and use of brain-inspired AI!

"A method for the ethical analysis of brain-inspired AI"
- See Publications

Capo Caccia Cognitive Neuromorphic Engineering Workshop - April 30 - May 14 - Info
Mihai A. Petrovici gave the talk "Energy-based approaches to neuronal dynamics and function" and Kevin Max "Latent Equilibrium and Phaseless Alignment Learning."

New contribution to a Preprint bringing together academia and industry members to define benchmarks for "brain-inspired" neuromorphic computing!

"NeuroBench: Advancing Neuromorphic Computing through Collaborative, Fair and Representative Benchmarking" - See Publication

Human Brain Project Summit 2023 - Mar. 28-31 - Info

Mihai A. Petrovici participated to the Session "The interplay of neuroscience and AI" and presented the talk: "Computing with physics: Learning from carbon and silicon".
Julian Göltz participated to the Session "Brain-inspired Robotics And Computing" and presented the talk "Fast and energy-efficient neuromorphic deep learning with first-spike times"

The Human Brain Project Podcast - Info
Mihai A. Petrovici talks about his research on intelligence, his experience of growing up in a family of physicists, and what he does for fun outside of work.

Listen to the podcast "The intersection of biological and artificial intelligence" on YouTube
and also on Spotify & Apple Podcasts

COSYNE Conference & Workshop - Mar. 9-14 - Info

Katharina Wilmes presented "Uncertainty-modulated prediction errors in cortical microcircuits" during the workshop session "Top-down interactions in the neocortex: Structure, function, plasticity and models."

Swiss Computational Neuroscience Retreat 2023 - Feb. 1-3 - Info

Katharina Wilmes presented
"Uncertainty-modulated prediction errors in cortical microcircuits".

New Preprint: (PAL), a bio-plausible method to learn efficient feedback weights!

"Learning efficient backprojections across cortical hierarchies in real time" - See Publication

Our group leader is the newly elected member of the EBRAINS STC!

The Science and Technology Committee of the EBRAINS Research Infrastructure will provide guidance and advice to the EBRAINS Board of Directors and Management Board on matters related to brain science, technology, and digital research tools and services.

Workshop: Energetics of computation in artificial & natural networks - Dec. 15/16 - Info

• Julian Göltz was invited to give the talk "From biology to silicon substrates: neural computation with physics".

Parisian Tour - Nov. 21-23

• Mihai was invited to give the talk "Computing With Physics: Intelligence In Carbon And Silicon" in Paris-Saclay University and the University of Paris to foster new collaborations.

Bernstein Conference - Sep. 13-16 - Info

Two talks

• Paul Haider presented "Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons". For more details, read the related article.
• Katharina Wilmes presented "Uncertainty-modulated prediction errors in cortical microcircuits".

And a poster

• Agnes Korcsak-Gorzo presented "Neural oscillations implement tempering in spiking neural networks".

EPFL: NeuroSpin school - Aug. 22-26 - Info

Mihai was invited to give the lecture "Introduction and state-of-the-art technologies used for neuromorphic computing".

Manfred-Stärk-Symposium 2022 - Lavaux - Sep. 11-12

Special symposium on theoretical neuroscience and neuromorphic computation dedicated to our generous benefactor Manfred Stärk and his foundation.

Nacht der Forschung 2022 - Info
We were delighted to be among the 750 researchers from the University of Bern who presented their research to tens of thousands of people.

Visitors to our stand got an insight into learning in neural networks and neuromorphic engineering by experiencing the dynamics of real biological neurons using our in-house Lu.i electronic neuron circuits. The public could also compete against an AI by playing the puzzle game Tic Tac Toe, also known as Noughts and Crosses.

Cellular and Molecular Neurobiology of Mental disease 2022 Meeting - Aug. 24-26

• Jakob Jordan was invited to give the talk "Dendritic integration as Bayes-optimal opinion pooling".

New Article in iScience! Approximating the ground states of quantum spin models.

"Variational learning of quantum ground states on spiking neuromorphic hardware".
See Publication - Blog post

CNS*2022 - Computational Neuroscience Meeting - July 16-20 - Info
Paul Haider gave two presentations: one at the main conference and the other at the workshop "Bio-inspired active AI workshop:"

"Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons" and
"Bio-inspired AI and AI-inspired biology."

New lab member (July)
Welcome to our new Research Assistant (Intership) Bennet Breier!

NEST Conference - June 23-24 - Info
Laura Kriener and Julian Göltz gave the talk:

"Fast and energy-efficient neuromorphic deep learning with first-spike times"

Lab Retreat in Kandersteg - June 7-10

Four days of science, brain-storming, metaphysical debates, sports, cooking and other chocolate testing. It was a lot of FUN!!

New Article in NICE Proceedings: An alternative to classic deep learning datasets!

"The Yin-Yang dataset". See Publication

New article in eLife: how natural-gradient descent enable efficient synaptic plasticity!

"Natural-gradient learning for spiking neurons".
See Publication

DEEP MINDS Podcast (in German) - Info

Laura Kriener and Julian Göltz talked about "AI Hardware of the Future: What is Neuromorphic Computing?"

From Cortical Microcircuits to Consciousness Symposium - Apr 11-13 - Info

• Nicolas Deperrois presented the poster "Learning cortical representations through perturbed and adversarial dreaming."
  

New contribution to eLife: Learning across three different global brain states!

"Learning cortical representations through perturbed and adversarial dreaming"
Publication - More on GANs

From Neuroscience to Artificially Intelligent Systems (NAISys) - Apr 5-9 - Info

• Paul Haider presented the poster "Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons."

NICE Conference - Mar 29-Apr 1 - Info
Three talks from Team NeuroTMA! Respectively, on March 29, 31 and April 1
Andreas Baumbach presented "Quantum many-body states: A novel neuromorphic application" and received the Best Early Researcher Presentation Award from NEUROTECH!!

• Laura Kriener presented "The Yin-Yang Dataset"
• Paul Haider presented "Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons"

Now in PCB: How cortical oscillations shape spike-based probabilistic inference in the brain!

"Cortical oscillations support sampling-based computations in spiking neural networks" - Publication

COSYNE Conference & Workshop - Mar 17-22 - Info

Paul Haider presented "Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons"

Happy to have contributed to shaping the vision of neuroscience in a HBP position paper! - Info

• "The coming decade of digital brain research - A vision for neuroscience at the intersection of technology and computing"

Swiss Computational Neuroscience Retreat 2022 - Mar. 2-4 - Info
Two Talks:

Jakob Jordan "Cortical representation learning & Bayes-optimal cue integration"
Paul Haider "Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons"

SIAM Conference on Parallel Processing for Scientific Computing - Feb 23-26 - Info

Andreas Baumbach presented "Computing with Biologically Inspired Dynamics"

6th HBP Student Conference on Interdisciplinary Brain Research - Feb. 22-25 - Info
Two poster presentations:

Ben von Hünerbein "Towards fully embedded biologically inspired deep learning on neuromorphic hardware"
Henrik Mettler "Evolving Neuronal Plasticity Rules using Cartesian Genetic Programming" - Recipient of the People’s Choice Best Abstract Award

New Preprint: a dataset of abstract paintings and nonfigurative art objects!

"DELAUNAY: a dataset of abstract art for psychophysical and machine learning research" - See Publication

New Article in SciPost Physics: our first demonstration of representing entangled quantum states on BrainScaleS!

"Spiking neuromorphic chip learns entangled quantum states" - See Publication