-
Effective Latent Differential Equation Models via Attention and Multiple Shooting
Authors:
Germán Abrevaya,
Mahta Ramezanian-Panahi,
Jean-Christophe Gagnon-Audet,
Pablo Polosecki,
Irina Rish,
Silvina Ponce Dawson,
Guillermo Cecchi,
Guillaume Dumas
Abstract:
Scientific Machine Learning (SciML) is a burgeoning field that synergistically combines domain-aware and interpretable models with agnostic machine learning techniques. In this work, we introduce GOKU-UI, an evolution of the SciML generative model GOKU-nets. GOKU-UI not only broadens the original model's spectrum to incorporate other classes of differential equations, such as Stochastic Differenti…
▽ More
Scientific Machine Learning (SciML) is a burgeoning field that synergistically combines domain-aware and interpretable models with agnostic machine learning techniques. In this work, we introduce GOKU-UI, an evolution of the SciML generative model GOKU-nets. GOKU-UI not only broadens the original model's spectrum to incorporate other classes of differential equations, such as Stochastic Differential Equations (SDEs), but also integrates attention mechanisms and a novel multiple shooting training strategy in the latent space. These modifications have led to a significant increase in its performance in both reconstruction and forecast tasks, as demonstrated by our evaluation of simulated and empirical data. Specifically, GOKU-UI outperformed all baseline models on synthetic datasets even with a training set 16-fold smaller, underscoring its remarkable data efficiency. Furthermore, when applied to empirical human brain data, while incorporating stochastic Stuart-Landau oscillators into its dynamical core, our proposed enhancements markedly increased the model's effectiveness in capturing complex brain dynamics. This augmented version not only surpassed all baseline methods in the reconstruction task, but also demonstrated lower prediction error of future brain activity up to 15 seconds ahead. By training GOKU-UI on resting state fMRI data, we encoded whole-brain dynamics into a latent representation, learning a low-dimensional dynamical system model that could offer insights into brain functionality and open avenues for practical applications such as the classification of mental states or psychiatric conditions. Ultimately, our research provides further impetus for the field of Scientific Machine Learning, showcasing the potential for advancements when established scientific insights are interwoven with modern machine learning.
△ Less
Submitted 14 September, 2023; v1 submitted 11 July, 2023;
originally announced July 2023.
-
Generative Models of Brain Dynamics -- A review
Authors:
Mahta Ramezanian Panahi,
Germán Abrevaya,
Jean-Christophe Gagnon-Audet,
Vikram Voleti,
Irina Rish,
Guillaume Dumas
Abstract:
The principled design and discovery of biologically- and physically-informed models of neuronal dynamics has been advancing since the mid-twentieth century. Recent developments in artificial intelligence (AI) have accelerated this progress. This review article gives a high-level overview of the approaches across different scales of organization and levels of abstraction. The studies covered in thi…
▽ More
The principled design and discovery of biologically- and physically-informed models of neuronal dynamics has been advancing since the mid-twentieth century. Recent developments in artificial intelligence (AI) have accelerated this progress. This review article gives a high-level overview of the approaches across different scales of organization and levels of abstraction. The studies covered in this paper include fundamental models in computational neuroscience, nonlinear dynamics, data-driven methods, as well as emergent practices. While not all of these models span the intersection of neuroscience, AI, and system dynamics, all of them do or can work in tandem as generative models, which, as we argue, provide superior properties for the analysis of neuroscientific data. We discuss the limitations and unique dynamical traits of brain data and the complementary need for hypothesis- and data-driven modeling. By way of conclusion, we present several hybrid generative models from recent literature in scientific machine learning, which can be efficiently deployed to yield interpretable models of neural dynamics.
△ Less
Submitted 23 December, 2021; v1 submitted 22 December, 2021;
originally announced December 2021.
-
Learning Nonlinear Brain Dynamics: van der Pol Meets LSTM
Authors:
German Abrevaya,
Irina Rish,
Aleksandr Y. Aravkin,
Guillermo Cecchi,
James Kozloski,
Pablo Polosecki,
Peng Zheng,
Silvina Ponce Dawson,
Juliana Rhee,
David Cox
Abstract:
Many real-world data sets, especially in biology, are produced by complex nonlinear dynamical systems. In this paper, we focus on brain calcium imaging (CaI) of different organisms (zebrafish and rat), aiming to build a model of joint activation dynamics in large neuronal populations, including the whole brain of zebrafish. We propose a new approach for capturing dynamics of temporal SVD component…
▽ More
Many real-world data sets, especially in biology, are produced by complex nonlinear dynamical systems. In this paper, we focus on brain calcium imaging (CaI) of different organisms (zebrafish and rat), aiming to build a model of joint activation dynamics in large neuronal populations, including the whole brain of zebrafish. We propose a new approach for capturing dynamics of temporal SVD components that uses the coupled (multivariate) van der Pol (VDP) oscillator, a nonlinear ordinary differential equation (ODE) model describing neural activity, with a new parameter estimation technique that combines variable projection optimization and stochastic search. We show that the approach successfully handles nonlinearities and hidden state variables in the coupled VDP. The approach is accurate, achieving 0.82 to 0.94 correlation between the actual and model-generated components, and interpretable, as VDP's coupling matrix reveals anatomically meaningful positive (excitatory) and negative (inhibitory) interactions across different brain subsystems corresponding to spatial SVD components. Moreover, VDP is comparable to (or sometimes better than) recurrent neural networks (LSTM) for (short-term) prediction of future brain activity; VDP needs less parameters to train, which was a plus on our small training data. Finally, the overall best predictive method, greatly outperforming both VDP and LSTM in short- and long-term predictive settings on both datasets, was the new hybrid VDP-LSTM approach that used VDP to simulate large domain-specific dataset for LSTM pretraining; note that simple LSTM data-augmentation via noisy versions of training data was much less effective.
△ Less
Submitted 20 July, 2019; v1 submitted 24 May, 2018;
originally announced May 2018.
