Feature and Language Selection in Temporal Symbolic Regression for Interpretable Air Quality Modelling

Abstract

Air quality modelling that relates meteorological, car traffic, and pollution data is a fundamental problem, approached in several different ways in the recent literature. In particular, a set of such data sampled at a specific location and during a specific period of time can be seen as a multivariate time series, and modelling the values of the pollutant concentrations can be seen as a multivariate temporal regression problem. In this paper, we propose a new method for symbolic multivariate temporal regression, and we apply it to several data sets that contain real air quality data from the city of Wrocław (Poland). Our experiments show that our approach is superior to classical, especially symbolic, ones, both in statistical performances and the interpretability of the results.

Artificial Intelligence in Physical Sciences: Symbolic Regression Trends and Perspectives

“…The latter, although extremely fast, produces non-interpretable equations [ 84 ]. In addition, there also exist Mixed-Integer Non-Linear Programming (MINLP) formulations [ 85 , 86 ], models that identify the SR problem as a linguistic [ 87 ], others that incorporate probabilistic features such as probabilistic framework [ 88 ] or probabilistic grammars [ 89 ], Bayesian approaches [ 90 ] and more [ 91 , 92 ].…”

“…However, from an environmentalist point of view, an energy surplus should not lead on an excessive use of available energy, by continuously increasing the income of generated power to adhere to our costly way of life; damage should be also minimized. Damage minimization can occur via energy consumption modelling [ 90 , 212 , 213 ], energy management [ 214 ], carbon emission studies [ 215 , 216 ], exhaust emission [ 217 ] and modelling of air quality [ 91 ]. All these applications can contribute towards a green energy transition and securing a healthier planet.…”

A Reinforcement Learning Approach to Domain-Knowledge Inclusion Using Grammar Guided Symbolic Regression

“…Toward this goal, the definition of explicit constraints can guarantee to produce only legal expressions with respect to the constraints. These constraints can be grammatical (Whigham et al, 1995) or ontological (Lucena-Sánchez et al, 2021). In Grammar-Guided Genetic Programming (G3P) (Whigham et al, 1995), also called Grammar-Based GP, a Context-Free Grammar (CFG) (Cremers and Ginsburg, 1975) is used to define constraint rules.…”

The voice of COVID-19: Breath and cough recording classification with temporal decision trees and random forests

“…Building on this idea, interval temporal logic random forests can be used to improve the performances of single interval temporal decision trees. Interval temporal forests follow the same principles as the propositional ones: a forest is a schema based on the idea of training different independent trees on different samples and different attributes; in the temporal case, moreover, they may also be trained on different interval relations (in [7] , the problem of selecting subsets of relations in the learning phase, treated as feature selection problem, has been studied).…”