Active learning of chemical reaction networksviaprobabilistic graphical models and Boolean reaction circuits

Abstract

Discerning networks of many reactions among multiple interconverting species is challenging. Here, we present a reaction network identification methodology. Our methodology enumerates all stoichiometrically and chemically feasible reactions and requires...

Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview

“…388 Probabilistic Graph Models provide such a framework, elements within categories being treated as random variables with conditional dependencies. 389 The old convolution concept 390 appears as a powerful grammar for recomposing mixture interactions, similar to the renormalization methods in electrodynamics. It can be applied to graphs through a weighted sum of node values, where the weights are determined by graph Laplacian eigenvalues corresponding to a range of spatial scales.…”

“…In fact, ANN graph representations of chemical mixtures demonstrate exceptional flexibility (up to hypergraphs) and expressivity for complex system analysis, visualization, interpretation, and transdisciplinary communication. − Combining them with probabilistic models and natural language processing may enable multifaceted deciphering of intricate relationships within a knowledge graph. , The concept of “fluid spatiality” enhances our understanding of networks by introducing dynamics into the connections, transitioning from “clocks” to “clouds” . Probabilistic Graph Models provide such a framework, elements within categories being treated as random variables with conditional dependencies …”

An active learning approach to model solid-electrolyte interphase formation in Li-ion batteries

“…The results for each class can be interpreted using D 2 [eV] 2 and Q 2 [eV]. In these tables, the reactions are divided into three categories: electrochemical (1-4), aggregation (5)(6)(7)(8)(9)(10)(11), and diffusion (12)(13)(14)(15), in order of their appearance on the table. The electrochemical reactions are divided into two categories: those that happen directly in contact with the electrode and those that happen through electron tunneling.…”

“…In this regard, machine learning techniques can be utilized. The active learning frameworks have been applied in various studies such as reaction networks, 15 finding free energies in chemical compounds, 16 and in Li-ion batteries to develop inter-atomic force fields. 17 In this work, we developed an active learning workflow.…”

LearnCK: mass conserving neural network reduction of chemistry and species of microkinetic models

“…Determining overall reaction rates from species rates is typically an ill-posed inverse problem (solving for the rhs of eqn ( 3)), and the set of overall reactions is often not unique (for recent developments see ref. 68).…”