Ismael Balafrej
@IsmaelBalafrej
Followers
45
Following
142
Statuses
22
PhD | Data scientist | Neuromorphic engineering
Canada
Joined February 2012
RT @DamienQuerlioz: Our self-powered memristor neural network! It harvests energy with a tiny solar cell, and adjusts its accuracy dependi…
0
7
0
Excited for our latest work to be published in @NatureComms 🎉
.@IsmaelBalafrej et al. @iemn_fr @genieUdeS propose a bio-inspired #artificialneuralnetwork #hardware able to self-adapt to solve new complex tasks, by autonomously connecting nodes using #electropolymerization Now out 👉@NatureComms
0
0
2
RT @hisspikeness: Announcing SNUFA 2023: 7-8 November (online). Your friendly focus meeting on spiking neural networks. Invited speakers: R…
0
15
0
RT @IOPneuromorphic: Now available as an Accepted Manuscript, free to read forever: Ahmad El Ferdaoussi, Jean Roua…
0
5
0
@imaneehm For recurrent neural networks, the story is a little bit different. With shorter timesteps you could have spikes from inhibitory neurons preventing other neurons from spiking. This is also a big motivation for smaller timesteps.
0
0
1
@imaneehm The effective decay over a fixed period of time is the same regardless of the choice of timestep when using an exact solution. The difference is actually during spike integration, where you "miss" spikes with a big dt. I'm not sure what you mean by no learning in ML.
1
0
0
Stay tuned for more neuromorphic, data science and general machine learning content ! And shoutout to @quarto_pub for making this experience effortless.
0
0
0
RT @MelikaPayvand: Call for paper for @icons_neuro 2023 is out! Deadline: April 15th Conference: August 1-3 We look forward to seeing y'al…
0
9
0
RT @IOPneuromorphic: Part of this paper from @GaspardGoupy Alexandre Juneau-Fecteau, @nikhil_garg20 @IsmaelBalafre…
0
3
0
@neuralreckoning @NicolasPerezNi1 But perhaps it is more accurate to scale the weights with half the leak ? Assuming a spike can happen with uniform distribution between 0..dt, then the average leakage of a spike would be exp(dt/2/tau) per clock cycle. Still, I think the refrac. period is the important part: 2/2
0
0
0
RT @C_Frenkel: I'm thrilled to announce that the 1st event of the tinyML neuromorphic initiative will take place online on September 27, 8:…
0
14
0
RT @neuralreckoning: Spiking neural network workshop announcement! We are delighted to invite you to SNUFA 22. Online, free workshop, Nove…
0
33
0
RT @MelikaPayvand: If you are at the @ISCAS2022 conference, come see us at the @JoinNeurotech booth and learn more about what we’re doing t…
0
8
0
RT @julie_grollier: We have several PhD and post-doc positions to work on learning through physics in hardware spintronic, memristive as w…
0
39
0
RT @neuromatch: Join NOW @neuralreckoning, @GunnarBlohm, @seanescola, @meganakpeters, @KordingLab, and @bradpwyble for our first #Neuromatc…
0
10
0
RT @IOPneuromorphic: Just published: @IsmaelBalafrej, Fabien Alibart and @JeanRouat from @USherbrooke @genieUdeS @sciencesUdeS look at a ne…
0
3
0