🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Computational neuroscience has lately had great success at modeling perception with ANNs - but it has been unclear if this approach translates to higher cognitive systems. We made some exciting progress in modeling human language processing biorxiv.org/content/10.110… #tweeprint 1/

We hope this work might also be of interest to others working on biologically inspired neural networks @neurograce Tony Zador Blake Richards Kenneth Stanley Jeff Clune Andrea Soltoggio Thomas Miconi Kording Lab 🦖 Stanford NeuroAI Lab hardmaru Jack Lindsey. Code will be published soon.

[1/2] Excited to share that iGibson 1.0 is accepted to #IROS2021 🤖 Cool features and scenes to develop and train robots for interactive tasks in large virtual environments: website: svl.stanford.edu/igibson/ Arxiv: arxiv.org/abs/2012.02924 Code: 👉 pip install "igibson==1.0.*" 👈

Large-language models like #GPT3 have shown tsunami impact in #AI world. We predict profound impact of these models in countless applications to come. 100+ researchers Stanford HAI have released a paper abt the present &future of these Foundation Models. arxiv.org/pdf/2108.07258…

(1/n) How to start a deep learning project? We use a remarkably streamlined step-by-step process to set up deep learning projects. At the same time, people who are new to deep learning tend to always make the same (avoidable) mistakes. Check out the thread below! 🧵

1/ Excited to share that our work on Deep Evolutionary Reinforcement Learning (DERL): a framework for large scale evolution of embodied agents in physically realistic environments is now published in Nature Communications Paper nature.com/articles/s4146… Video youtube.com/watch?v=zltE0w…

1/ People study medial entorhinal cortex (MEC) because of its key role in navigation and memory. In our new paper we build neural network models that explain the full diversity of neural responses in MEC. To appear as a #NeurIPS2021 Spotlight! Tweetprint below 👇🧠

Happy to share my Neuromatch #NMC4 flash talk on this work! youtube.com/watch?v=VEE1Ct…

Do AI models understand everyday physical phenomena like we do, or are they missing key elements of human judgment? Check out #Physion, our #NeurIPS2021 benchmark for comparing physical prediction in models and humans, to find out! Paper, Dataset, Talk: bit.ly/3Evq3r5

1/ Interested in how task-driven neural networks help us understand the diverse cell types in medial entorhinal cortex (MEC), a brain area that plays a key role in navigation & memory? Stop by #NeurIPS2021 poster session 6 tomorrow (8:30 am PT, Spot F0)! tinyurl.com/4vkwzj86

1/3 We release our ImageNet pretrained Recurrent CNN models, which currently best explain neural dynamics & temporally varying visual behaviors. Ready to be used with 1 line of code! Models: tinyurl.com/ms88azrv Paper (to appear in Neural Computation): tinyurl.com/2p8kcka7

Yesterday I successfully defended my PhD dissertation, "A Goal-Driven Approach to Systems Neuroscience". Especially grateful to Daniel Yamins and Surya Ganguli for advising me throughout this journey! Defense Talk: youtu.be/WED5GPKEv4Q Thesis: anayebi.github.io/files/thesis.p…

1/8 Can we use embodied AI to gain insight into *why* neural systems are as they are? In previous work👇, we demonstrated that a contrastive unsupervised objective substantially outperforms supervised object categorization at generating networks that predict mouse visual cortex.

8/8 We think answering these Why questions is one of most productive uses for task-driven deep neural networks in neuroscience. More details found here: biorxiv.org/content/10.110…

1/15 This is an important point worth underscoring. As I’ll elaborate here below -- there is actually a lot of shared perspective between critiques of NeuroAI with the main considerations of those who practice it. It also leads to some new directions that I'll note!🧵👇

9/15 Alongside theoretical considerations, an empirical criterion for picking a (model, metric) pair could be if it enables better neural population control, compared to alternatives. A good scientific theory should help control the phenomena it seeks to explain.

11/15 Finally, the discussion of hyperparameters has come up most recently in the context of MEC models. Actually, one main result in our NeurIPS '21 paper was that pure path integration alone was *not* sufficient to match responses in MEC. (Figs 2 & 3 in biorxiv.org/content/10.110…)

If you're attending #cosyne2023, stop by our Poster III-002 👇 this Saturday at 8:30 pm on "Mouse visual cortex as a limited resource system that self-learns an ecologically-general representation"! w/ Nathan Kong Chengxu Zhuang Justin Gardner Anthony M. Norcia Daniel Yamins

At CosyneMeeting, our poster at 3-022 will discuss how to measure similarity between DNN model activations and biological neural responses. Please stop by if you're interested! w/ Javier Sagastuy* (* equal co-authors) Aran Nayebi Rosa Cao Daniel Yamins #cosyne2023

Excited to share our work on Mouse Visual Cortex is now published in @ploscompbiol: journals.plos.org/ploscompbiol/a… #ploscompbiolauthor We also release our SOTA pretrained models & our Pytorch library of SSL methods so others can build their own models easily: github.com/neuroailab/mou…