arXiv subjects

Georgios Stratis

Publications and source records attributed to Georgios Stratis.

5 recordsLinked to original sources

arXiv2026

RTS Smoother-Guided Learning of Physics-Based Neural Differential Models

Ordinary differential equations (ODEs) are widely used to model dynamical systems in physics, biology, neuroscience, and physiology, but in many applications some equations of the dynamics are unknown and only a subset of the state variables are measured. We propose a hybrid neural--physics framework in which the known components of the ODE are kept explicit and the missing components are represented by a neural network. The proposed method consists of two stages where we alternate between state and parameter estimation and iterate until a predetermined criterion is met. Specifically, in the first step, we treat the model parameters as being known and we infer the latent states from the available measurements using a Rauch--Tung--Striebel (RTS) smoother. In the second stage, we treat the smoothed trajectories as being known and use them to estimate the neural networks' parameters through backpropagation. We evaluate the method on benchmark systems spanning linear, nonlinear, and stiff dynamics under partial state observation. Across these settings, the proposed method learns missing ODE components from incomplete measurements while exploiting and retaining interpretable mechanistic structure and improving latent-state reconstruction and long-horizon prediction.

cs.LG↗

arXiv2026

Temporal Point Process Modeling of Aggressive Behavior Onset in Psychiatric Inpatient Youths with Autism

Aggressive behavior, including aggression towards others and self-injury, occurs in up to 80% of children and adolescents with autism, making it a leading cause of behavioral health referrals and a major driver of healthcare costs. Predicting when autistic youth will exhibit aggression can be challenging due to their communication difficulties. Many are minimally verbal or have poor emotional insight. Recent advances in Machine Learning and wearable biosensing demonstrate the ability to predict aggression within a limited future window (typically one to three minutes) in autistic individuals. However, existing works don't estimate aggression onset probability or the expected number of aggression onsets over longer periods, nor do they provide interpretable insights into onset dynamics. To address these limitations, we apply Temporal Point Processes (TPPs) - particularly self-exciting Hawkes processes - to model the timing of aggressive behavior onsets in psychiatric inpatient autistic youth. We benchmark several TPP models by evaluating their goodness-of-fit and predictive metrics. Our results demonstrate that self-exciting TPPs more accurately captures the irregular and clustered nature of aggression onsets, especially compared to traditional Poisson models. These incipient findings suggest that TPPs can provide interpretable, probabilistic forecasts of aggression onset along a time continuum, supporting future clinical decision-making and preemptive intervention.

stat.AP↗

arXiv2025

Exploring Theory-Laden Observations in the Brain Basis of Emotional Experience

In the science of emotion, it is widely assumed that folk emotion categories form a biological and psychological typology, and studies are routinely designed and analyzed to identify emotion-specific patterns. This approach shapes the observations that studies report, ultimately reinforcing the assumption that guided the investigation. Here, we reanalyzed data from one such typologically-guided study that reported mappings between individual brain patterns and group-averaged ratings of 34 emotion categories. Our reanalysis was guided by an alternative view of emotion categories as populations of variable, situated instances, and which predicts a priori that there will be significant variation in brain patterns within a category across instances. Correspondingly, our analysis made minimal assumptions about the structure of the variance present in the data. As predicted, we did not observe the original mappings and instead observed significant variation across individuals. These findings demonstrate how starting assumptions can ultimately impact scientific conclusions and suggest that a hypothesis must be supported using multiple analytic methods before it is taken seriously.

cs.LG↗

arXiv2024

One-half reflected entropy is not a lower bound for entanglement of purification

In recent work, Akers et al. proved that the entanglement of purification $E_p(A:B)$ is bounded below by half of the $q$-Rényi reflected entropy $S_R^{(q)}(A:B)$ for all $q\geq2$, showing that $E_p(A:B) = \frac{1}{2} S_R^{(q)}(A:B)$ for a class of random tensor network states. Naturally, the authors raise the question of whether a similar bound holds at $q = 1$. Our work answers that question in the negative by finding explicit counter-examples, which we arrive at through numerical optimization. Nevertheless, this result does not preclude the possibility that restricted sets of states, such as CFT states with semi-classical gravity duals, could obey the bound in question.

quant-ph↗

arXiv2022

Sample generation for the spin-fermion model using neural networks

Quantum Monte-Carlo simulations of hybrid quantum-classical models such as the double exchange Hamiltonian require calculating the density of states of the quantum degrees of freedom at every step. Unfortunately, the computational complexity of exact diagonalization grows $ \mathcal{O} (N^3)$ as a function of the system's size $ N $, making it prohibitively expensive for any realistic system. We consider leveraging data-driven methods, namely neural networks, to replace the exact diagonalization step in order to speed up sample generation. We explore a model that learns the free energy for each spin configuration and a second one that learns the Hamiltonian's eigenvalues. We implement data augmentation by taking advantage of the Hamiltonian's symmetries to artificially enlarge our training set and benchmark the different models by evaluating several thermodynamic quantities. While all models considered here perform exceedingly well in the one-dimensional case, only the neural network that outputs the eigenvalues is able to capture the right behavior in two dimensions. The simplicity of the architecture we use in conjunction with the model agnostic form of the neural networks can enable fast sample generation without the need of a researcher's intervention.

physics.comp-ph↗