
New preprint out! Probabilistic Forecasting via Autoregressive Flow Matching. arxiv.org/html/2503.10375v1

The Artificial Cognitive Systems lab @DondersInst headed by @marcelge investigates the computational principles that govern natural and artificial intelligence.

New preprint out! Probabilistic Forecasting via Autoregressive Flow Matching. arxiv.org/html/2503.10375v1

Our new preprint "PAM: Predictive attention mechanism for neural decoding of visual perception" introduces a novel approach that learns output queries. Beneficial if queries are n/a, as in neural decoding! w/ @lynnle_ai @artcogsys @yagmurgucluturk @umuguc biorxiv.org/content/10.1101/…

Discovering dynamic symbolic policies with genetic programming. The evolved policies: -consist of interpretable dynamics equations -offer transparency in the latent state -robust under partial observability -generalize to different task settings arxiv.org/abs/2406.02765

I'm excited to share our new preprint 'Subspace Node Pruning'! We present a novel perspective on pruning computational units from pre-trained deep networks with minimal performance loss. (0/10)

Train RNNs efficiently without gradients! Our learning algorithms (ANP & DANP) offer a robust solution by - Requiring only forward passes - Avoiding vanishing gradients - Matching BP (BPTT) performance - Enabling neuromorphic hardware compatibility arxiv.org/pdf/2405.08967v1

[1/8] Our paper "Brain2GAN: Feature-disentangled neural encoding and decoding of visual perception in the primate brain" is published! @PLOSCompBiol #brainreading 🧠📖 Full paper: doi.org/10.1371/journal.pcbi…

As AI adoption becomes more widespread, so do concerns about its energy consumption. Our latest paper proposes a method to train AI models more efficiently. The method reduces energy use by more than half when training a ResNet-18 model on ImageNet. arxiv.org/pdf/2405.02385

Check our paper to see how a simple bistable model can capture the essence of epileptic dynamics. Networks of such bistable models can be even more powerful in interpreting real data. Joint work with @A_ElGazzarr @DaniSBassett @Fabiopas82 @marcelge
A new publication from DBI2 is available🥳! "Analytical Characterization of Epileptic Dynamics in a Bistable System" arxiv.org/abs/2404.03409


Excited to share our perspective paper on Universal differential equations for neuroscience. arxiv.org/abs/2403.14510. In this manuscript, we argue for UDEs as a unifying framework to integrate mathematical modeling with machine learning in neuroscience.


Come join us! Postdoctoral research position in efficient machine learning as part of the DBI2 research consortium: ru.nl/node/67534

We are very happy to announce that the DBI2 website is finally live!🥳 dbi2.nl/. Please visit our website to find out about our team and research activities!

New paper out! Advancements in spiking neural network communication and synchronization techniques for event-driven neuromorphic systems sciencedirect.com/science/ar…

Unlocking speech for paralyzed individuals using Brain-Computer Interface Technology and AI ru.nl/donders/news/news/@138… via @Radboud_Uni

Beyond excited: our manuscript on PhenoScore is out in @NatureGenet! By combining facial recognition technology with phenotypic data analyses, we recognize 37/40 investigated syndromes and even establish new phenotypic subgroups for #ADNP! tinyurl.com/yaent4yw