Nishanth Anand

Continual learning should be viewed through the lens of the stability-plasticity dilemma, and solving it requires rethinking learning architectures. I argue why in my first continual learning blog: itsnva7.substack.com/p/continual-le…

Montreal deep tech scene is getting hot!! Many recent hires of Cohere, Mistral, Periodic Labs, Poolside are all based in Montreal. And now, AMI will have an office here 🔥 It's a no-brainer, though. @Mila_Quebec has the highest concentration of deep learning expertise with interdisciplinary connections. Thanks to recent US regulation changes on immigration, no more brain drain! Let's build more in Canada!

Don't forget to tune in tomorrow, Tuesday, March 3rd for a session with @itsNVA7 focused on "The permanent and transient framework for continual reinforcement learning." Learn more: cohere.com/events/cohere-…

@bronzeagepapi One solution that I proposed: openreview.net/pdf?id=5XfxEQ2…

@itsNVA7 Recommend these, they’re great mental models arxiv.org/abs/2512.24695 arxiv.org/abs/2504.13173

Continual learning should be viewed through the lens of the stability-plasticity dilemma, and solving it requires rethinking learning architectures. I argue why in my first continual learning blog: itsnva7.substack.com/p/continual-le…

@MaheshwariWork Nice idea! I believe the separation should also be along representations to enable instance-based learning or learning from rare-events, which NNs can’t do. You might find my works interesting: arxiv.org/abs/2312.11669 openreview.net/pdf?id=5XfxEQ2…

@itsNVA7 Agreed! I decouple plasticity and stability through a dual network architecture. Check this out: arxiv.org/abs/2512.01034 (Recently accepted at AAMAS)

@bronzeagepapi Hahah! Science is slow, and I have a long life ahead of me to figure it out.

@itsNVA7 Indeed

@bronzeagepapi Possibly! There are many more solutions that fit the desiderata.

@itsNVA7 Associative memory?

This post is inspired by numerous discussions with my Ph.D. Supervisor, Doina Precup. And thanks to @khurram and my friends for their valuable comments on the initial draft.

This strategy ensures rapid learning from a rare event by leveraging the transient system, which also shields the permanent system from abrupt changes demanded by the raw online experience, thereby preserving previously learned knowledge.

I am excited to share the past 5+ years of my PhD work on continual reinforcement learning at @Cohere_Labs on March 3rd! Doina and I spent significant time and thought developing this framework; we believe this holds the key to continual learning.

David Silver's new $4bn company, Ineffable Intelligence, will fulfil the promise of the Era of Experience. youtube.com/watch?v=zzXyPG… ft.com/content/dffe72…

Since AI has been added to sinus surgery - "Two (patients) suffered strokes after surgeons accidentally damaged carotid arteries while the system allegedly misinformed them about where their instruments were inside patients' heads."

8 years ago, I sat in Doina’s RL class as an enthusiastic MSc student. This winter, I’m co-teaching that same class with her at McGill 👨‍🏫 Life comes full circle 🔁

Had a chance to attend this talk at our group meeting. Very exciting work!!

Thrilled to share our new #NeurIPS2025 paper done at @GoogleDeepMind, Plasticity as the Mirror of Empowerment We prove every agent faces a trade-off between its capacity to adapt (plasticity) and its capacity to steer (empowerment) Paper: david-abel.github.io/plasticity.pdf 🧵🧵🧵👇

The Dwarkesh/Andrej interview is worth watching. Like many others in the field, my introduction to deep learning was Andrej’s CS231n. In this era when many are involved in wishful thinking driven by simple pattern matching (e.g., extrapolating scaling laws without nuance), it’s refreshing to hear an influential voice that is tethered to reality. One clarification for the podcast is that when Andrej says humans don’t use reinforcement learning, he is really saying humans don't use returns as learning targets. His example of LLMs struggling to learn to solve math problems from outcome-based rewards also elucidates the problem with learning directly from returns. Fortunately for RL, this exact problem is solved by temporal difference (TD) learning. All sample-efficient RL algorithms that show human-like learning (e.g., sample-efficient learning on Atari, and our work on learning from experience directly on a robot) rely on TD learning. Now Andrej is not primarily an RL person; he is looking at RL through the lens of LLMs these days, and all RL done in LLMs uses returns as targets, so it’s understandable that he is assuming that RL is all about learning from observed returns. But this assumption leads him to the incorrect conclusion that we need process-based dense rewards for RL to work. If you embrace TD learning, then you don't necessarily need a dense reward. Once you have learned a value function that encodes useful knowledge about the world, you can learn on the fly in the absence of rewards, just like humans and animals. This is possible because in TD learning there is no difference between learning from an unexpected reward and learning from an unexpected change in perceived value.

Multi-Agent RL fails in real life. Agents cooperating to solve tasks remains a utopia. -No scalable algorithms for general-sum games. -In a simple apple-harvesting game, PPO agents overharvest and ruin bushes. Advantage Alignment (ICLR 2025 Oral📢) is a huge step forward. 1/n

@MartinKlissarov @GoogleDeepMind @egrefen Congratulations Martin! 🎉

Thrilled to share I've joined @GoogleDeepMind in London! Grateful to work with the brilliant @egrefen & a great team towards open-ended autonomous assistants. To celebrate, I took my daughter to see the best view in London—she made sure I saw nothing. Guess I need an assistant!