Comparison of Decision Tree Based Classification Strategies to Detect External Chemical Stimuli from Raw and Filtered Plant Electrical Response
📝 Abstract
Plants monitor their surrounding environment and control their physiological functions by producing an electrical response. We recorded electrical signals from different plants by exposing them to Sodium Chloride (NaCl), Ozone (O3) and Sulfuric Acid (H2SO4) under laboratory conditions. After applying pre-processing techniques such as filtering and drift removal, we extracted few statistical features from the acquired plant electrical signals. Using these features, combined with different classification algorithms, we used a decision tree based multi-class classification strategy to identify the three different external chemical stimuli. We here present our exploration to obtain the optimum set of ranked feature and classifier combination that can separate a particular chemical stimulus from the incoming stream of plant electrical signals. The paper also reports an exhaustive comparison of similar feature based classification using the filtered and the raw plant signals, containing the high frequency stochastic part and also the low frequency trends present in it, as two different cases for feature extraction. The work, presented in this paper opens up new possibilities for using plant electrical signals to monitor and detect other environmental stimuli apart from NaCl, O3 and H2SO4 in future.
💡 Analysis
Plants monitor their surrounding environment and control their physiological functions by producing an electrical response. We recorded electrical signals from different plants by exposing them to Sodium Chloride (NaCl), Ozone (O3) and Sulfuric Acid (H2SO4) under laboratory conditions. After applying pre-processing techniques such as filtering and drift removal, we extracted few statistical features from the acquired plant electrical signals. Using these features, combined with different classification algorithms, we used a decision tree based multi-class classification strategy to identify the three different external chemical stimuli. We here present our exploration to obtain the optimum set of ranked feature and classifier combination that can separate a particular chemical stimulus from the incoming stream of plant electrical signals. The paper also reports an exhaustive comparison of similar feature based classification using the filtered and the raw plant signals, containing the high frequency stochastic part and also the low frequency trends present in it, as two different cases for feature extraction. The work, presented in this paper opens up new possibilities for using plant electrical signals to monitor and detect other environmental stimuli apart from NaCl, O3 and H2SO4 in future.
📄 Content
Sensors and Actuators B: Chemical 1
Comparison of Decision Tree Based Classification Strategies to Detect External Chemical Stimuli from Raw and Filtered Plant Electrical Response
Shre Kumar Chatterjee1, Saptarshi Das1, Koushik Maharatna1, Elisa Masi2, Luisa Santopolo2, Ilaria Colzi2, Stefano Mancuso2 and Andrea Vitaletti3,4 1 School of Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ, UK 2 Department of Agri-food Production and Environmental Science, University of Florence, FIorence, Italy 3 DIAG, Sapienza University of Rome, via Ariosto 25, 00185 Rome, Italy Email: s.das@soton.ac.uk, sd2a11@ecs.soton.ac.uk (S. Das*) Phone: +44(0)7448572598, Fax: 02380 593045 Abstract Plants monitor their surrounding environment and control their physiological functions by producing an electrical response. We recorded electrical signals from different plants by exposing them to Sodium Chloride (NaCl), Ozone (O3) and Sulfuric Acid (H2SO4) under laboratory conditions. After applying pre-processing techniques such as filtering and drift removal, we extracted few statistical features from the acquired plant electrical signals. Using these features, combined with different classification algorithms, we used a decision tree based multi-class classification strategy to identify the three different external chemical stimuli. We here present our exploration to obtain the optimum set of ranked feature and classifier combination that can separate a particular chemical stimulus from the incoming stream of plant electrical signals. The paper also reports an exhaustive comparison of similar feature based classification using the filtered and the raw plant signals, containing the high frequency stochastic part and also the low frequency trends present in it, as two different cases for feature extraction. The work, presented in this paper opens up new possibilities for using plant electrical signals to monitor and detect other environmental stimuli apart from NaCl, O3 and H2SO4 in future. Index Terms Decision tree, multiclass classification, discriminant analysis, Mahalanobis distance classifier, statistical features, plant electrical signal processing, time series analysis I. Introduction Plants, such as Mimosa pudica (Touch-me-not) and Helianthus annuus (Sunflower), show some form of physical changes due to external stimuli in the form of touch and sunlight Sensors and Actuators B: Chemical 2
respectively [1]. The wilting of general plants due to dry environmental conditions is also commonly found. For many years, researchers have tried to establish the relationship between these reactions of the plants and the surrounding environmental conditions [1]. It has been found that the underlying phenomenon behind this is the plant electrophysiological mechanism which may be traced in the electrical response of the plant to the external stimulus [1]. These electrical signals, which control various physiological functions in the plants, hold useful information about the external stimulus (which causes the electrical signal in the plant) contained within its deterministic and stochastic parts to different extents. Analysis using low frequency (trend) part of the plant electrical signal to study the external chemical or light stimulus has been reported in Chatterjee et al. [2], [3]. Also, other studies on plant electrical signal processing have been reported in [4]–[14], in particular use of classification techniques to find out the applied external stimuli, through various statistical features computed from the recorded plant electrical signal, was reported first in [2]. Since the statistical features in [2] were extracted from raw plant signals (with low frequency trends or drifts), a background (pre- stimulus) information subtraction method was adopted in the classification process to focus only on the incremental values in each feature due to the application of the stimulus. In this paper, we initially focus on the information contained in the stochastic part of the plant electrical signals by applying a high pass filtering on the raw signals to remove the inconsistent trends or drifts. We also used raw signals with the trends (using the background information subtraction method as reported in [2]) to show a comparative analysis between the classification performance of the filtered and raw plant signals. Thus, we here explore, if there is any improvement in the classification process while using only the detrended random part rather than the raw signal containing small local fluctuations superimposed on relatively larger change in the trends. In order to develop a classification strategy for detecting the external chemical stimuli, here we used 15 features out of which 11 features have been reported in [2]. In addition to these 11 features, four additional statistical features have been explored along with independent testing of the classifi
This content is AI-processed based on ArXiv data.