AI for dynamical systems: the architecture that incorporates 1) a block of DNN for embeddings (feature engineering) and 2) a block of Koopman Operator based predictor/controller is emerging as a promising one in terms of accuracy and lean computation. Kolmogorov Arnold Networks

The combination of sampling and Koopman theory means we can use linear control theory, such as LQR, when applying RNNs to controlled dynamical systems. Compared to RC, our method performs comparably and without a very time and computationally consuming hyperparameter search. 4/4

By sampling the hidden layers, we avoid backpropagation-through-time (BPTT). This alleviates the problems with BPTT, such as exploding/vanishing gradients. We explore RNN in the light of Koopman theory, which allows us to prove convergence of our method. 3/4

This drastically decreases computational time while remaining competitive on performance, which we demonstrate on several known DS-uncontrolled and controlled systems- and weather data. Plus, the outer-layer linear maps provide a connection between RNNs and Koopman operators. 2/4

Excited to share our new paper, https://t.co/PPWnfrkyV2, with my amazing collaborators at @TU_Muenchen! We construct all RNN weights and biases without gradient-based methods, using random feature networks and Koopman operator theory. 1/4

Symbolic dynamics bridges from #DynamicalSystems to computation/ #AI! In our #NeurIPS2024 (@NeurIPSConf) paper we present a new network architecture, Almost-Linear RNNs, that finds most parsimonious piecewise-linear representations of DS from data: https://t.co/sePJ4WHGkg 1/4

Just wanted to stop by & say: We have 2 new accepted #NeurIPS2024 papers: 1) @brenner_manuel , Hemmer, @Zahra__Monfared, DD: Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction --> *this takes DSR to a new level!*, details to follow

امیرعلی عسگری: طلا امیرحسین فرخنده‌فر و امیررضا درستی: نقره پارسا فرج‌پور: برنز تیم ملی ایران در المپیاد جهانی کامپیوتر ۲۰۲۴ مصر ✌️

Data-driven linearization (DDL) explains dynamic mode decomposition (DMD) as a first-order, approximate linearization of the dynamics on a dominant spectral submanifold. DDL then refines this linearization to higher orders with increased accuracy. https://t.co/PD9kRuqkmO

تیم ملی ایران در نخستین المپیاد جهانی هوش مصنوعی ✌️

The pro Koopman modeling tool is soon to be released!!! #AI #ML #Koopman #TimeSeries

How Will You Use PyKMD? Share Your Ideas! We’re excited to announce the launch of PyKMD, our cutting-edge software for time series data analysis! Now, we want to hear from you: how do you plan to use PyKMD in your work or research? #DataScience #Machinelearning #AI #koopman

If you are around for #ASSC27 and want to hear about decision-theoretic approaches to perceptual multistability, come to my talk!

Here is the link for open positions: https://t.co/m0YcaKl5ak

Honored to receive prestigious #BMBF funding in #AI to establish my junior research group! Returning from @TU_Muenchen to @UniHeidelberg in Sept 2024. Seeking 1 postdoc and 2 PhD students. Apply by July 20 if you're passionate about the #Mathematical_Foundations_of_AI

Honored to receive prestigious #BMBF funding in #AI to establish my junior research group! Returning from @TU_Muenchen to @UniHeidelberg in Sept 2024. Seeking 1 Postdoc, and 2 PhD students. Apply by July 20 if you're passionate about the #Mathematical_Foundations_of_AI

Interested in investigating how neuronal dynamics supports reinforcement learning? Join us! 📢 PhD position in the Brain Dynamics Laboratory at @SantAnnaPisa, Italy. #RNN #ReinforcementLearning #compneuro Deadline: at 13:00 of the 29 July 2024. https://t.co/sXyzlA3Jbb

🚀We're seeking a passionate postdoc to join our team and work on #DynamicalSystems, #GenerativeAI, and #Neuroimaging, focused on #mentalhealth. Start date: Oct. 2024 or soon after. Interested? Contact me and share with your network! #Postdoc #Research #MentalHealthResearch

Evidence is emerging LLM's are linear representations of sequences https://t.co/yxgrS22ZN0 . I conjectured this in

Happy Birthday Dear Maryam! Celebrating Maryam Mirzakhani's birthday anniversary with a special gift—a mug featuring her portrait! 🎉 Truly one of the best presents ever. Remembering her remarkable contributions to mathematics and her inspiring legacy. #MaryamMirzakhani