13) Really fun project with Julius Vetter @fel_p8 @_rdgao @jakhmack🙌. Thanks to @mackelab and reviewers for comments, Willet et al. & Churchland and Cunningham for making their datasets publicly available.🙏🙏.

12) There's much more! Come visit us at our poster if you are at NeurIPS, tomorrow Wednesday 11 am or check out our paper ➡️ :. Poster details:.East Exhibit Hall A-C #4010.Wed 11 Dec 11 a.m. PST — 2 p.m. PST.

Blazing-fast image creation – using just your voice. Try Grok Imagine.

11) LDNS is particularly promising for heterogeneous datasets without trial structure, which pose challenges for many LVMs. LDNS successfully mimicked cortical data during attempted speech—a challenging task due to varying trial lengths.

10) Colorful latents are just so nice to look at 🙃, so we were glad to see that the LDNS latent space preserves behavioral information. Both latents and PCs thereof reflect the reach direction of reaches used for conditioning.

9) LDNS allows for flexible conditioning on behavioral variables. Diffusion models conditioned on either reach direction or velocity trajectories produce neural activity samples that are consistent with the queried behavior.

8) To increase the realism of generated spikes even further, we demonstrate how to equip LDNS with more expressive autoregressive observation models. (this can be applied to any LVM trained with Poisson log-likelihood!)

7) We then moved to a classic monkey reach task and show that LDNS samples are indistinguishable to the human eye from real cortical data and accurately capture population level and single neuron statistics.

6) We validate that LDNS does what it’s supposed to on simulated spiking data. LDNS perfectly captured firing rates & underlying dynamics and can length-generalize—producing faithful samples of 16 times the original training length.

5) But how does LDNS work?. The AE first maps spikes to time-aligned latents, which allows training flexible (un)conditional diffusion models on smoothly varying latents, circumventing the issue of diffusion models acting on discrete values.

4) Latent Diffusion for Neural Spiking data to the rescue ⛑️. LDNS combines 1) a regularized S4-based autoencoder (AE) with 2) diffusion in latent space, and can model diverse neural spiking data. Here we consider 3 very different tasks:

3) ✅Diffusion models, on the other hand, can produce faithful samples across many domains and allow for flexible conditioning on external variables. ❌Yet dealing with discrete data such as spikes poses a challenge for them.

2) ✅LVMs have been highly successful in neuroscience for inferring low dimensional dynamics underlying neural population activity,.❌but are often not designed with the goal of producing high-fidelity samples. 🧠🙉🐭.

1) With our @NeurIPSConf poster happening tomorrow, it's about time to introduce our Spotlight paper 🔦, co-lead with @_Jaivardhan_ :. Latent Diffusion for Neural Spiking data (LDNS), a latent variable model (LVM) which addresses 3 goals simultaneously:

RT @ai_rhet: Join us in #Tübingen for the #PersuasiveAlgorithms conference on the rhetoric of #GenAI!. 📍𝗠𝗣𝗜 𝗳𝗼𝗿 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝘁 𝗦𝘆𝘀𝘁𝗲𝗺𝘀, 𝗦𝗽𝗲𝗺𝗮𝗻….

RT @MPI_IS: Researching #AI #ComputerVision #MachineLearning #Robotics #HCI? Join our elite doctoral program - a partnership with MPI-IS, @….

RT @mackelab: We are looking to hire multiple PhD students on (1) deep learning for mechanistic models of neural computation, (2) simulatio….

RT @_rdgao: Want a tool that uses ML to generate REALLY good fake brain recordings?. You're getting one. Julius' paper on diffusion models….

RT @uni_tue: Zwei Attempto-Preise für neurowissenschaftliche Arbeiten: Matthias Baumann und Roxana Zeraati werden für herausragende Studien….

RT @tuewiml: Last Friday was a blast at the second TWiML Workshop! 🎉 Huge thanks to all the speakers and the participants. Special thanks t….

RT @tuewiml: Only one more day to go! 🎉.Have you checked out the program yet?.We can’t wait to see you tomorrow!💫.@….