White-box Induction From SVM Models: Explainable AI with Logic Programming

We focus on the problem of inducing logic programs that explain models learned by the support vector machine (SVM) algorithm. The top-down sequential covering inductive logic programming (ILP) algorithms (e.g., FOIL) apply hill-climbing search using heuristics from information theory. A major issue with this class of algorithms is getting stuck in local optima. In our new approach, however, the data-dependent hill-climbing search is replaced with a model-dependent search where a globally optimal SVM model is trained first, then the algorithm looks into support vectors as the most influential data points in the model, and induces a clause that would cover the support vector and points that are most similar to that support vector. Instead of defining a fixed hypothesis search space, our algorithm makes use of SHAP, an example-specific interpreter in explainable AI, to determine a relevant set of features. This approach yields an algorithm that captures the SVM model’s underlying logic and outperforms other ILP algorithms in terms of the number of induced clauses and classification evaluation metrics.

Designing Weather Based Crop Insurance Payout Estimation Based on Agro-Meteorological Data using Machine Learning Techniques

Data mining techniques have been extensively used to mine up-to-date information from agricultural databases. In Agriculture, the Loss Assessment and Estimation in Crop insurance can be done on various factors like yield-based, crop-health based and weather-based variations. Weather-based variations are taken into account to design the insurance payout classifier model for the selected crop within the selected agricultural blocks of Tamilnadu. Then the weather attributes that undergone feature selection are given as input to the model with the rule-based classification algorithm implementing the neighboring approach with a sequential covering strategy named as CBKNN-PAYRULE which is statistically higher than other state-of-the-art rule-based classification algorithms. This model is proposed to classify the agricultural blocks based on the Area-wise Assessment of adverse temperature for the groundnut crop from their nearest neighbor. Then By combining the classified neighboring approach with the threshold factors the Rule-based classifier is done to generate the rules to estimate the insurance payout value as per policymakers for the selected agricultural blocks. Then decision-making techniques are applied to predict the insurance with the possibility of product basis risk, which covers the deviations in weather indices with the risk profile factors for the notified agricultural blocks for the specified crop. Thus the proposed technique can support the simultaneous prediction of the insurance payout to be paid in case of adverse weather factors of the selected crop for five districts with high accuracy and the correlation analysis of weather factors with the payout concerning to each district is also made. The Experimental results show that the proposed work enhances the accuracy in insurance payout prediction of the groundnut crop of the selected districts

Study of population dynamics of soybean semi-looper Gesonia gemmaSwinhoe by using rule induction model in Maharashtra, India

Gesonia gemma Swinhoe (1885) is a grey semi-looper and it has emerged as a serious threat to the soybean crop. This defoliator causes heavy damage to the crop in the form of loss in grain weight. Gesonia gemma population dynamics was studied in various districts of Maharashtra. Sequential covering algorithm (CN2 rule induction) has been proposed for rule induction model to generate a list of classification rules with target feature (G. gemma population) and the independent abiotic features. The classification rules have exhibited more accuracy and showed that maximum temperature and humidity with less number of rainy days has influenced the population of Gesonia gemma in Maharashtra. Hence, this rule induction model can be used to study the collected evidence for prediction and it will be helpful to the farmers to take necessary pest control strategy.