RB-GAT: A Text Classification Model Based on RoBERTa-BiGRU with Graph ATtention Network.

With the development of deep learning, several graph neural network (GNN)-based approaches have been utilized for text classification. However, GNNs encounter challenges when capturing contextual text information within a document sequence. To address this, a novel text classification model, RB-GAT, is proposed by combining RoBERTa-BiGRU embedding and a multi-head Graph ATtention Network (GAT). First, the pre-trained RoBERTa model is exploited to learn word and text embeddings in different contexts. Second, the Bidirectional Gated Recurrent Unit (BiGRU) is employed to capture long-term dependencies and bidirectional sentence information from the text context. Next, the multi-head graph attention network is applied to analyze this information, which serves as a node feature for the document. Finally, the classification results are generated through a Softmax layer. Experimental results on five benchmark datasets demonstrate that our method can achieve an accuracy of 71.48%, 98.45%, 80.32%, 90.84%, and 95.67% on Ohsumed, R8, MR, 20NG and R52, respectively, which is superior to the existing nine text classification approaches.

Nanomaterials for Targeted Delivery of Agrochemicals by an All-in-One Combination Strategy and Deep Learning.

The development of modern agriculture has prompted the greater input of herbicides, insecticides, and fertilizers. However, precision release and targeted delivery of these agrochemicals still remain a challenge. Here, a pesticide-fertilizer all-in-one combination (PFAC) strategy and deep learning are employed to form a system for controlled and targeted delivery of agrochemicals. This system mainly consists of three components: (1) hollow mesoporous silica (HMS), to encapsulate herbicides and phase-change material; (2) polydopamine (PDA) coating, to provide a photothermal effect; and (3) a zeolitic imidazolate framework (ZIF8), to provide micronutrient Zn2+ and encapsulate insecticides. Results show that the PFAC at concentration of 5 mg mL-1 reaches the phase transition temperature of 1-tetradecanol (37.5 °C) after 5 min of near-infrared (NIR) irradiation (800 nm, 0.5 W cm-2). The data of corn and weed are collected and relayed to deep learning algorithms for model building to realize object detection and further targeted weeding. In-field treatment results indicated that the growth of chicory herb was significantly inhibited when treated with the PFAC compared with the blank group after 24 h under NIR irradiation for 2 h. This system combines agrochemical innovation and artificial intelligence technology, achieves synergistic effects of weeding and insecticide and nutrient supply, and will potentially achieve precision and sustainable agriculture.