The rabbit breeding industry exhibits vast economic potential and growth opportunities. Nevertheless, the ineffective prediction of environmental conditions in rabbit houses often leads to the... Show More

The rabbit breeding industry exhibits vast economic potential and growth opportunities. Nevertheless, the ineffective prediction of environmental conditions in rabbit houses often leads to the spread of infectious diseases, causing illness and death among rabbits. This paper presents a multi-parameter predictive model for environmental conditions such as temperature, humidity, illumination, CO2 concentration, NH3 concentration, and dust conditions in rabbit houses. The model adeptly distinguishes between day and night forecasts, thereby improving the adaptive adjustment of environmental data trends. Importantly, the model encapsulates multi-parameter environmental forecasting to heighten precision, given the high degree of interrelation among parameters. The model's performance is assessed through RMSE, MAE, and MAPE metrics, yielding values of 0.018, 0.031, and 6.31% respectively in predicting rabbit house environmental factors. Experimentally juxtaposed with Bert, Seq2seq, and conventional transformer models, the method demonstrates superior performance.

In recent years, deep learning has been widely used as an efficient prediction algorithm. However, this algorithm has strict requirements on the size of training samples. If there are not enough... Show More

In recent years, deep learning has been widely used as an efficient prediction algorithm. However, this algorithm has strict requirements on the size of training samples. If there are not enough samples to train the network, it is difficult to achieve the desired effect. In view of the lack of training samples, this article proposes a deep learning prediction model integrating migration learning and applies it to flood forecasting. The model uses random forest algorithm to extract the flood characteristics, and then uses the transfer learning strategy to fine-tune the parameters of the model based on the model trained with similar reservoir data; and is used for the target reservoir flood prediction. Based on the calculation results, an autoregressive algorithm is used to intelligently correct the error of the prediction results. A series of experimental results show that our proposed method is significantly superior to other classical methods in prediction accuracy.

Uncertain information in the securities market exhibits fuzziness. In this article, expected returns and liquidity are considered as trapezoidal fuzzy numbers. The possibility mean and mean absolute... Show More

Uncertain information in the securities market exhibits fuzziness. In this article, expected returns and liquidity are considered as trapezoidal fuzzy numbers. The possibility mean and mean absolute deviation of expected returns represent the returns and risks of securities assets, while the possibility mean of expected turnover represents the liquidity of securities assets. Taking into account practical constraints such as cardinality and transaction costs, this article establishes a fuzzy portfolio model with cardinality constraints and solves it using the differential evolution algorithm. Finally, using fuzzy c-means clustering algorithm, 12 stocks are selected as empirical samples to provide numerical calculation examples. At the same time, fuzzy c-means clustering algorithm is used to cluster the stock yield data and analyse the stock data comprehensively and accurately, which provides a reference for establishing an effective portfolio.

Accurately identifying rumor information is crucial for efficient information assessment. Pretrained Language Models (PLMs) are trained on large text data, understanding and generating human-like... Show More

Accurately identifying rumor information is crucial for efficient information assessment. Pretrained Language Models (PLMs) are trained on large text data, understanding and generating human-like language patterns, and fine-tuning for specific NLP tasks. However, adjusting PLMs typically requires modifications for specific tasks, creating a gap between pretraining and task execution. Prompt learning reduces this gap. We introduce “Prompt Learning for Rumor Detection” (PLRD) based on T5, which utilizes T5-generated prompt templates to transform the detection task into a prompt-driven learning framework. Through precise prompt guidance, PLRD leverages model knowledge, enhancing rumor detection capabilities, especially in data-scarce scenarios. Experimental validation on Weibo and Twitter datasets confirms PLRD's superiority over existing methods, particularly in scenarios with limited data. Comparative analysis against state-of-the-art methods highlights PLRD's competitiveness and advancement in rumor detection.

The recommender system can be viewed as a matrix completion problem, which aims to predict unknown values within a matrix. Solutions to this problem are categorized into two approaches: transductive... Show More

The recommender system can be viewed as a matrix completion problem, which aims to predict unknown values within a matrix. Solutions to this problem are categorized into two approaches: transductive and inductive reasoning. In transductive reasoning, the model cannot be applied to new cases unseen during training. In contrast, IGMC, the state-of-the-art inductive algorithm, only requires subgraphs for target users and items, without needing any other content information. While the absence of a requirement for content information simplifies the model and enhances transferability to new tasks, incorporating content information could still improve the model's performance. In this article, the authors introduce Hi-GMC, a hybrid version of the IGMC model that incorporates content information alongside users and items. They present a novel graph model to encapsulate the side information related to users and items and develop a learning method based on graph neural networks. This proposed method achieves state-of-the-art performance on the MovieLens-100K dataset for both warm and cold start scenarios.

With the development of smart education, gaining insights into students' understanding during the learning process is crucial in teaching. However, traditional knowledge tracking methods face... Show More

With the development of smart education, gaining insights into students' understanding during the learning process is crucial in teaching. However, traditional knowledge tracking methods face challenges in capturing the intricate relationships between problems and knowledge points, as well as students' temporal learning changes. Therefore, we design a knowledge tracking model based on a graph temporal fusion network. Firstly, we construct the structure of the question and knowledge skill graph. Then, we design a knowledge graph encoder layer to capture the complex relationships between questions and knowledge skills. Next, we apply a sequential information extraction layer to dynamically model the outputs of each layer in the upper network over time, capturing students' knowledge changes at different time steps. Finally, we use a dynamic attention aggregation network to learn node information at different levels and time sequences. Experimental results on three datasets demonstrate the effectiveness of our method.