Search

Plant diseases are a major threat to food security and economic prosperity around the globe. Deep learning models based on Convolution Neural Network (CNN) have shown promising results in dealing with plant disease detection tasks. However, according to the No Free Lunch Theorem, no single model is suitable for all cases. Moreover, the vast diversity of plant diseases makes the model selection process time and resource extensive, using exhaustive search. This work proposes a meta-learning-based framework that recommends top-n suitable models for an unseen plant disease detection dataset using the prior evaluations of benchmark models on plant disease detection tasks. Rank-Biased Overlap (RBO) is used to evaluate the efficacy of the proposed framework by evaluating actual rankings with respect to the predicted rankings. Extensive comparative experiments are carried out with different configurations of meta-extractors and meta-learners. The results obtained demonstrate that the probe network trained for 10 epochs (termed as "intermediate stage") along with standard deviation as meta-extractor and Support Vector Regressor as the meta-learner outperforms the rest with average RBO scores of 0.76, 0.73 and 0.75 for Top-5, Top-3 and Top-1 recommendations, respectively. Overall, this paper presents a viable substitute for the exhaustive search process carried out for choosing the best deep learning model for plant disease detection scenario, leading to better resource utilization and faster implementation procedure. • Meta-learning framework to recommend models for plant disease identification task. • Meta-dataset created using 13 benchmark CNN models on 24 different species of plants. • RBO score used to evaluate combinations of meta-extractors and meta-learners. [ABSTRACT FROM AUTHOR]

• A new hybrid deep learning based framework for photovoltaic power forecasting is proposed. • The framework integrates long short term memory layer with vanishing time series gradient and maximal overlap discrete wavelet transform series. • MODWT is implemented using a multiresolution pyramidal hierarchical decomposition technique. • The proposed method outperforms previous models and establishes its efficacy even for longer intervals. During this era of the energy crisis, when the non-renewable sources are rapidly diminishing, efforts are being taken to utilize renewable sources predominantly. This manuscript presents a hybrid deep learning framework using long short term memory (LSTM) Layer with vanishing time series gradient and maximal overlap discrete wavelet transform (MODWT) model for photovoltaic (PV) power forecasting through time series decomposition. The proposed framework is implemented on the dataset collected from Yulara Solar System, Australia. During the experimental evaluation, obtained results demonstrate short term temporal dependence of PV power forecasting on solar power magnitudes as well as weather conditions. Moreover, the proposed model outperforms existing state-of-the-art models in terms of mean average percentage error (MAPE) by 14.17%, 3.01%, and 16.49% for 1 day, 10 days, and 1 month, respectively, establishing its efficacy even for longer intervals. Proposed Ensemble Model divided into three stages viz. Time series decomposition and reconstruction, forecasting phase, and weighted aggregation of predicted results [Display omitted] [ABSTRACT FROM AUTHOR]