Mathematics & kindness.

We wrote a little #NeuroAI piece about in-context learning & neural dynamics vs. continual learning & plasticity, both mechanisms to flexibly adapt to changing environments:.We relate this to non-stationary rule learning w rapid jumps. Feedback welcome!.

Fantastic work by Florian Bähner, Hazem Toutounji, Tzvetan Popovand & many others - I'm just the person advertising!.

How do animals learn new rules? By systematically testing diff. behavioral strategies, guided by selective attn. to rule-relevant cues: Akin to in-context learning in AI, strategy selection depends on the animals' "training set" (prior experience).

RT @russo_eleon: Into population dynamics? Coming to #CNS2025 but not quite ready to head home?. Come join us! at the Symposium on "Neural….

We dive a bit into the reasons why current time series FMs not trained for DS reconstruction fail, and conclude that a DS perspective on time series forecasting & models may help to advance the #TimeSeriesAnalysis field. (6/6)

Remarkably, DynaMix not only generalizes zero-shot to novel DS, but it can even generalize to new initial conditions and regions of state space not covered by the in-context information. (5/6)

And no, it’s neither based on Transformers nor Mamba – it’s a new type of mixture-of-experts architecture based on the recently introduced AL-RNN (, specifically trained for DS reconstruction. #AI.(4/6)

It often even outperforms TS FMs on forecasting diverse empirical time series, like weather or traffic, typically used to train TS FMs. This is surprising, cos DynaMix’ training corpus consists *solely* of simulated limit cycles & chaotic systems, no empirical data at all!.(3/6)

Unlike TS FMs, DynaMix exhibits #ZeroShotLearning of long-term stats of unseen DS, incl. attrac. geom. & PS, w/o *any* re-training, just from a context signal. It does so with only 0.1% of the parameters of Chronos & 10x faster inference times than the closest competitor. (2/6)

Can time series #FoundationModels like Chronos zero-shot generalize to unseen #DynamicalSystems (DS)?. No, they cannot. But *DynaMix* can, the first FM based on principles of DS reconstruction, capturing the long-term evolution of out-of-domain DS: (1/6)

Our revised #ICLR2025 paper & code for a foundation model architecture for dynamical systems is now online: incl. add. examples of how this may be used for identifying drivers (control par.) of non-stationary processes. And please switch platform!.

Transfer & few-shot learning for dynamical systems . our paper just got accepted for #ICLR2025 @iclr_conf !.Thread below; strongly updated version will be available soon . and don't forget to move to bsky!.

Periodic reminder that MSE or explained var. are not good stats for assessing quality of dynamical systems reconstructions, cos of exponential trajectory divergence in chaotic systems. And please follow us at !

Can really recommend this excellent talk by Christoph Bergmeir at yesterday … on the inability of Transformer- & LLM-based recent time series models to beat even simple baselines *if you do the stats and testing right*! A lesson in careful stat. eval.

2) A scalable generative model for dynamical system reconstruction from neuroimaging data. @GeorgiaKoppe.

Don't miss out on our 2 #NeurIPS2024 papers on dynamical systems reconstruction today & tomorrow:. 1) Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction .

RT @russo_eleon: 📢PhD position @BristolUni (with Ross Purple and @seanfw, UK) and joint supervision @SantAnnaPisa (with @russo_eleon, IT) o….

Happy our team is among this years’ recipients of the Samsung Global Research Outreach Awards!. We will take #DynamicalSystems reconstruction to the next level, large-scale – looking forward to the collaboration with the Samsung team!.

Hey everyone,.Please follow us on bluesky:. it's a wonderful place!.