Automatic classification of plant electrophysiological responses to environmental stimuli using machine learning and interval arithmetic

Abstract

In plants, there are different types of electrical signals involving changes in membrane potentials that could encode electrical information related to physiological states when plants are stimulated by different environmental conditions. A previous study analyzing traits of the dynamics of whole plant low-voltage electrical showed, for instance, that some specific frequencies that can be observed on plants growing under undisturbed conditions disappear after stress-like environments, such as cold, low light and osmotic stimuli. In this paper, we propose to test different methods of automatic classification in order to identify when different environmental cues cause specific changes in the electrical signals of plants. In order to verify such hypothesis, we used machine learning algorithms (Artificial Neural Networks, Convolutional Neural Network, Optimum-Path Forest, k-Nearest Neighbors and Support Vector Machine) together Interval Arithmetic. The results indicated that Interval Arithmetic and supervised classifiers are more suitable than deep learning techniques, showing promising results towards such research area.

Evaluation of genotoxicity after acute and chronic exposure to 2,4-dichlorophenoxyacetic acid herbicide (2,4-D) in rodents using machine learning algorithms

“…k-NN is a simple and effective technique, and it is known for making use of all samples, which were used for training as a prototype through the input parameter k. The clas-sification of a sample is achieved considering the nearest k training samples. For the value of k = 1, the test sample ends up being classified according to the class of the training sample that is closest to it (Pereira et al, 2018).…”

Electrophysiological assessment of plant status outside a Faraday cage using supervised machine learning

“…This is a significant finding. Other studies have explored strategies for the classification algorithms of low-voltage variations (microvolt) 33 or raw non-stationary 34 plant electrical signals reaching an accuracy of 84.4% and 73.7%, respectively. Further investigations are required to verify whether this model can be generalized to other plant species.…”

“…First, from the raw signal, in steps of 5 minutes we repetitively took seven samples of different window sizes, namely 15 s, 30 s, 1 min, 2 min, 5 min, 10 min and 30 min. Then, in each window, we extracted 26 features: simple statistical features (min, max, mean, variance, skewness, kurtosis and interquartile range), Hjorth parameters (mobility and complexity), Generalized Hurst exponent, Wavelet entropy (Shannon and logarithmic) 34 and the estimation of the color of the noise (white, pink, brown, blue and purple) 33 . We also perform a wavelet decomposition, of order 1, 4 and 8, on the windowed signals and took the corresponding min, max and average value as additional features.…”

Automatic Determination of the Parameters of Electrical Signals and Functional Responses of Plants Using the Wavelet Transformation Method

“…It is necessary to note the applications of other approaches for automatic signal processing in plants, including use of different types of classifiers [63], interval arithmetic [67], Fourier transform [64], vector computing [65,67], and multiple neural network learning algorithms [67,68]. The methods based on combination of several approaches, including those using WT, seem to be the most effective [63].…”