Retrieval-Augmented Generation (RAG) plays a vital role in enhancing large language models (LLMs) by enabling them to leverage external knowledge, thereby reducing hallucinations and improving the accuracy of responses. However, current RAG systems lack the capability to adjust retrieval strategies according to the complexity of queries. This inflexibility often leads to poor performance. To address this, we propose a novel Adaptive Retrieval-Augmented Generation framework. It employs an improved multi-armed bandit algorithm as a classifier for predicting query complexity. Treating each retrieval method as an ”arm”, the algorithm balances exploration and exploitation via Relative Performance. Experiments on multiple single-hop and multi-hop datasets demonstrate robust results: it outperforms traditional static RAG systems in both accuracy and efficiency, with significant gains in handling complex multi-hop queries while not adding unnecessary computational overhead for simple single-hop ones.

Driven by the development of cloud computing and artificial intelligence, architecture has dramatically improved in terms of flexibility and scalability in software development. Therefore, it is increasingly being used to build large-scale applications for agile development. However, along with the technology heterogeneity, the dynamics of running instances, and the complexity of service dependencies, fault localization is extraordinarily difficult. In this paper, we present MicroMILTS, a microservice fault location method based on mutual information and an LSTM Autoencoder. MicroMILTS first uses BIRCH for anomaly detection based on the analysis of the performance metrics data correlated to microservice anomalies. Once anomalies are detected, a service dependency property graph is constructed based on the real-time microservice invocation relationships and the reconstructed deviations of performance metrics with the LSTM Autoencoder. Next, MicroMILTS dynamically updates the weight of each node in the service dependency property graph. Then, a PageRank-based random walk is applied for further ranking root causes. Finally, a Sock-shop microservice system is built on the Huawei Cloud to evaluate the performance of MicroMILTS. The experiment shows that MicroMILTS achieves a good root cause location result, with 90.4 % in precision and 91.6% in mean average precision, outperforming state-of-the-art methods.

Since most existing saliency detection models are not suitable for the condition that the salient objects are near at the image border, the authors propose a saliency detection approach based on adaptive background template (SCB) despite of the position of the salient objects. First, a selection strategy is presented to establish the adaptive background template by removing the potential saliency superpixels from the image border regions, and the initial saliency map is obtained. Second, a propagation mechanism based on K ‐means algorithm is designed for maintaining the neighbourhood coherence of the above saliency map. Finally, a new spatial prior is presented to integrate the saliency detection results by aggregating two complementary measures such as image centre preference and the background template exclusion. Comprehensive evaluations on six benchmark datasets indicate that the authors’ method outperforms other state‐of‐the‐art approaches. In addition, a new dataset containing 300 challenging images is constructed for evaluating the performance of various salient object detection methods.

Co-saliency detection, an emerging research area in saliency detection, aims to extract the common saliency from the multi images. The extracted co-saliency map has been utilized in various applications, such as in co-segmentation, co-recognition and so on. With the rapid development of image acquisition technology, the original digital images are becoming more and more clearly. The existing co-saliency detection methods processing these images need enormous computer memory along with high computational complexity. These limitations made it hard to satisfy the demand of real-time user interaction. This paper proposes a fast co-saliency detection method based on the image block partition and sparse feature extraction method (BSFCoS). Firstly, the images are divided into several uniform blocks, and the low-level features are extracted from Lab and RGB color spaces. In order to maintain the characteristics of the original images and reduce the number of feature points as well as possible, Truncated Power for sparse principal components method are employed to extract sparse features. Furthermore, K-Means method is adopted to cluster the extracted sparse features, and calculate the three salient feature weights. Finally, the co-saliency map was acquired from the feature fusion of the saliency map for single image and multi images. The proposed method has been tested and simulated on two benchmark datasets: Co-saliency Pairs and CMU Cornell iCoseg datasets. Compared with the existing co-saliency methods, BSFCoS has a significant running time improvement in multi images processing while ensuring detection results. Lastly, the co-segmentation method based on BSFCoS is also given and has a better co-segmentation performance.

Cluster analysis has become a popular method for gene expression data, which can be used for the diagnosis of diseases accurately and rapidly through the class label. However, more attributes and less samples of gene expression data will produce a mass of redundant or disturbed information, resulting in the decline of the accuracy of the direct clustering acting on high dimensional data. Principal Component Analysis (PCA) is a classical method for dimension reduction which can transform high dimension data into low space. The shortcoming of PCA is the lack of strong interpretation because the loadings have no characteristic of sparsity. In this paper, a sparse PCA method based on Truncated Power, which can minimizes the cardinality of loadings as well as maximizes the percentage explained variances of principal components (PCs), was applied into the feature extraction method for gene expression, then the sparse PCs was fed into K-means process for clustering. Finally, the experimental results on three typical gene datasets verify that the sparse gene data can improve the efficiency and accuracy on clustering analysis.