Julian JN
LightNews LightNews
banner
julian-jn.bsky.social
Julian JN
@julian-jn.bsky.social
12 followers 17 following 14 posts
Neuro-inspired AI researcher at CitAI, City University of London
Posts Replies Media Videos
julian-jn.bsky.social
🌟 Weekend Read: Advancing AI with Hebbian Learning! 🌟

Explore our latest research on integrating Hebbian learning into CNNs for more biologically realistic unsupervised models!

🔗 arxiv.org/abs/2501.17266

🔗 github.com/Julian-JN/Ad...

🌐 Follow @emp1.bsky.social & @citai.bsky.social for updates!
Advancing the Biological Plausibility and Efficacy of Hebbian Convolutional Neural Networks
The research presented in this paper advances the integration of Hebbian learning into Convolutional Neural Networks (CNNs) for image processing, systematically exploring different architectures to bu...
arxiv.org
February 7, 2025 at 11:06 AM
julian-jn.bsky.social
🧠Excited to share my published paper [...]
Brain on cold meds: "Look at my published paper!"
Reality: "Sir, this is an arxiv preprint"

Still thrilled to share our Hebbian CNN work though! 🤒✨

*Corrected version:*
🧠 Excited to share our new preprint [...]. Check it out: arxiv.org/abs/2501.17266
January 31, 2025 at 10:06 AM
julian-jn.bsky.social
1. 🧠Excited to share my published paper "Advancing the Biological Plausibility and Efficacy of Hebbian Convolutional Neural Networks" with @emp1.bsky.social. Available from: arxiv.org/abs/2501.17266
January 30, 2025 at 6:07 PM
Reposted by Julian JN
emp1.bsky.social
An advance of a paper that will soon be available in ArXiv. All merit to my wonderful student @julian-jn.bsky.social !
I hope you enjoy it.

We present an optimal architecture that significantly enhances recent research aimed at integrating Hebbian learning with competition mechanisms and CNNs.
➡️
Advancing the Biological Plausibility and Efficacy of Hebbian Convolutional Neural Networks by Julian Jimenez-Nimo and Esther Mondragón The research presented in this paper advances the integration of Hebbian learning into Convolutional Neural Networks (CNNs) for image processing, systematically exploring different architectures to build an optimal configuration, adhering to biological tenability. Hebbian learning operates on local unsupervised neural information to form feature representations, providing an alternative to the popular but arguably biologically implausible and computationally intensive backpropagation learning algorithm. The suggested optimal architecture significantly enhances recent research aimed at integrating Hebbian learning with competition mechanisms and CNNs, expanding their representational capabilities by incorporating hard Winner-Takes-All (WTA) competition, Gaussian lateral inhibition mechanisms and Bienenstock–Cooper–Munro (BCM) learning rule in a single model. The resulting model achieved 76\% classification accuracy on CIFAR-10, rivalling its end-to-end backpropagation variant (77\%) and critically surpassing the state-of-the-art hard-WTA performance in CNNs of the same network depth (64.6\%) by 11.4\%. Moreover, results showed clear indications of sparse hierarchical learning through increasingly complex and abstract receptive fields. In summary, our implementation enhances both the performance and the generalisability of the learnt representations and constitutes a crucial step towards more biologically realistic artificial neural networks.
January 6, 2025 at 6:34 PM