Artificial Intelligence Review最新文献

Applying deep learning (DL) for annotating surgical instruments in robot-assisted minimally invasive surgeries (MIS) represents a significant advancement in surgical technology. This systematic review examines 48 studies that utilize advanced DL methods and architectures. These sophisticated DL models have shown notable improvements in the precision and efficiency of detecting and segmenting surgical tools. The enhanced capabilities of these models support various clinical applications, including real-time intraoperative guidance, comprehensive postoperative evaluations, and objective assessments of surgical skills. By accurately identifying and segmenting surgical instruments in video data, DL models provide detailed feedback to surgeons, thereby improving surgical outcomes and reducing complication risks. Furthermore, the application of DL in surgical education is transformative. The review underscores the significant impact of DL on improving the accuracy of skill assessments and the overall quality of surgical training programs. However, implementing DL in surgical tool detection and segmentation faces challenges, such as the need for large, accurately annotated datasets to train these models effectively. The manual annotation process is labor-intensive and time-consuming, posing a significant bottleneck. Future research should focus on automating the detection and segmentation process and enhancing the robustness of DL models against environmental variations. Expanding the application of DL models across various surgical specialties will be essential to fully realize this technology’s potential. Integrating DL with other emerging technologies, such as augmented reality (AR), also offers promising opportunities to further enhance the precision and efficacy of surgical procedures.

Complex networks are commonly observed in several real-world areas, such as social, biological, and technical systems, where they exhibit complicated patterns of connectedness and organised clusters. These networks have intricate topological characteristics that frequently elude conventional characterization. Link prediction in complex networks, like data flow in telecommunications networks, protein interactions in biological systems, and social media interactions on platforms like Facebook, etc., is an essential element of network analytics and presents fresh research challenges. Consequently, there is a growing emphasis in research on creating new link prediction methods for different network applications. This survey investigates several strategies related to link prediction, ranging from feature extraction based to feature learning based techniques, with a specific focus on their utilisation in dynamic and developing network topologies. Furthermore, this paper emphasises on a wide variety of feature learning techniques that go beyond basic feature extraction and matrix factorization. It includes advanced learning-based algorithms and neural network techniques specifically designed for link prediction. The study also presents evaluation results of different link prediction techniques on homogeneous and heterogeneous network datasets, and provides a thorough examination of existing methods and potential areas for further investigation.

Cancer gene expression data has the characteristics of high-dimensional, multi-text and multi-classification. The problem of cancer subtype diagnosis can be solved by selecting the most representative and predictive genes from a large number of gene expression data. Feature selection technology can effectively reduce the dimension of data, which helps analyze the information on cancer gene expression data. A multi-strategy fusion binary sea-horse optimizer based on Gaussian transfer function (GOG-MBSHO) is proposed to solve the feature selection problem of cancer gene expression data. Firstly, the multi-strategy includes golden sine strategy, hippo escape strategy and multiple inertia weight strategies. The sea-horse optimizer with the golden sine strategy does not disrupt the structure of the original algorithm. Embedding the golden sine strategy within the spiral motion of the sea-horse optimizer enhances the movement of the algorithm and improves its global exploration and local exploitation capabilities. The hippo escape strategy is introduced for random selection, which avoids the algorithm from falling into local optima, increases the search diversity, and improves the optimization accuracy of the algorithm. The advantage of multiple inertial weight strategies is that dynamic exploitation and exploration can be carried out to accelerate the convergence speed and improve the performance of the algorithm. Then, the effectiveness of multi-strategy fusion was demonstrated by 15 UCI datasets. The simulation results show that the proposed Gaussian transfer function is better than the commonly used S-type and V-type transfer functions, which can improve the classification accuracy, effectively reduce the number of features, and obtain better fitness value. Finally, comparing with other binary swarm intelligent optimization algorithms on 15 cancer gene expression datasets, it is proved that the proposed GOG1-MBSHO has great advantages in the feature selection of cancer gene expression data.

Measuring Efficiency in neural network system development is an open research problem. This paper presents an experimental framework to measure the training efficiency of a neural architecture. To demonstrate our approach, we analyze the training efficiency of Convolutional Neural Networks and Bayesian equivalents on the MNIST and CIFAR-10 tasks. Our results show that training efficiency decays as training progresses and varies across different stopping criteria for a given neural model and learning task. We also find a non-linear relationship between training stopping criteria, training Efficiency, model size, and training Efficiency. Furthermore, we illustrate the potential confounding effects of overtraining on measuring the training efficiency of a neural architecture. Regarding relative training efficiency across different architectures, our results indicate that CNNs are more efficient than BCNNs on both datasets. More generally, as a learning task becomes more complex, the relative difference in training efficiency between different architectures becomes more pronounced.

Using (p,q,r-) fractional fuzzy sets ((p,q,r-) FFS) to demonstrate the stability of cryptocurrencies is considered due to the complex and volatile nature of cryptocurrency markets, where traditional models may fall short in capturing nuances and uncertainties. (p,q,r-) FFS provides a flexible framework for modeling cryptocurrency stability by accommodating imprecise data, multidimensional analysis of various market factors, and adaptability to the unique characteristics of the cryptocurrency space, potentially offering a more comprehensive understanding of the factors influencing stability. Existing studies have explored Picture Fuzzy Sets and Spherical Fuzzy Sets, built on membership, neutrality, and non-membership grades. However, these sets can’t reach the maximum value (equal to (1)) due to grade constraints. For example, when considering (wp =(h,langle text{0.9,0.8,1.0}rangle left|hin Hright.)), these sets fall short. This is obvious when a decision-maker possesses complete confidence in an alternative, they have the option to assign a value of 1 as the assessment score for that alternative. This signifies that they harbor no doubts or uncertainties regarding the chosen option. To address this, (p,q,r-) Fractional Fuzzy Sets ((p,q,r-) FFSs) are introduced, using new parameters (p), (q), and (r). These parameters abide by (p),(qge 1) and (r) as the least common multiple of (p) and (q). We establish operational laws for (p,q,r-) FFSs. Based on these operational laws, we proposed a series of aggregation operators (AOs) to aggregate the information in context of (p,q,r-) fractional fuzzy numbers. Furthermore, we constructed a novel multi-criteria group decision-making (MCGDM) method to deal with real-world decision-making problems. A numerical example is provided to demonstrate the proposed approach.

Heart disease is the most significant health problem around the world. Thus, it emphasizes the need for accurate and efficient predictive models for early diagnosis. This study proposes an innovative approach integrating the Electric Eel Foraging Optimization Algorithm (EEFOA) with the Random Forest (RF) algorithm for classifying heart disease prediction. EEFOA draws inspiration from the foraging behaviour of electric eels, a bio-inspired optimization framework capable of effectively exploring complex solutions. The objective is to improve the predictive performance of heart disease diagnosis by integrating optimization and Machine learning methodologies. The experiment uses a heart disease dataset comprising clinical and demographic features of at-risk individuals. Subsequently, EEFOA was applied to optimize the features of the dataset and classification using the RF algorithm, thereby enhancing its predictive performance. The results demonstrate that the Electric Eel Foraging Optimization Algorithm Random Forest (EEFOARF) model outperforms traditional RF and other state-of-the-art classifiers in terms of predictive accuracy, sensitivity, specificity, precision, and Log_Loss, achieving remarkable scores of 96.59%, 95.15%, 98.04%, 98%, and 0.1179, respectively. The proposed methodology has the potential to make a significant contribution, thereby reducing morbidity and mortality rates.

The task of predicting entities and relations in Temporal Knowledge Graph (TKG) extrapolation is crucial and has been studied extensively. Mainstream algorithms, such as Gated Recurrent Unit (GRU) models, primarily focus on encoding historical factual features within TKGs, often neglecting the importance of incorporating entities and relational features during decoding. This bias ultimately leads to loss of detail and inadequate prediction accuracy during the inference process. To address this issue, a novel ChronoBridge framework is proposed that features a dual mechanism of a chronological node encoder and a bridged feature fusion decoder. Specifically, the chronological node encoder employs an advanced recursive neural network with an enhanced GRU in an autoregressive manner to model historical KG sequences, thereby accurately capturing entity changes over time and significantly enhancing the model’s ability to identify and encode temporal patterns of facts across the timeline. Meanwhile, the bridged feature fusion decoder utilizes a new variant of GRU and a multilayer perception mechanism during the prediction phase to extract entity and relation features and fuse them for inference, thereby strengthening the reasoning capabilities of the model for future events. Testing on three standard datasets showed significant improvements, with a 25.21% increase in MRR accuracy and a 39.38% enhancement in relation inference. This advancement not only improves the understanding of temporal evolution in knowledge graphs but also sets a foundation for future research and applications of TKG reasoning.

Given the pressing need for explainability in Machine Learning systems, the studies on counterfactual explanations have gained significant interest. This research delves into this timely problem cast in a unique context of relational systems described by fuzzy relational equations. We develop a comprehensive solution to the counterfactual problems encountered in this setting, which is a novel contribution to the field. An underlying optimization problem is formulated, and its gradient-based solution is constructed. We demonstrate that the non-uniqueness of the derived solution is conveniently formalized and quantified by admitting a result coming in the form of information granules of a higher type, namely type-2 or interval-valued fuzzy set. The construction of the solution in this format is realized by invoking the principle of justifiable granularity, another innovative aspect of our research. We also discuss ways of designing fuzzy relations and elaborate on methods of carrying out counterfactual explanations in rule-based models. Illustrative examples are included to present the performance of the method and interpret the obtained results.

With the explosive growth of user-generated content, multi-view learning has become a rapidly growing direction in pattern recognition and data analysis areas. Due to the significant application value of multi-view learning, there has been a continuous emergence of research based on machine learning methods and traditional deep learning paradigms. The core challenge in multi-view learning lies in harnessing both consistent and complementary information to forge a unified, comprehensive representation. However, many multi-view learning tasks are based on graph-structured data, making existing methods unable to effectively mine the information contained in the input multiple data sources. Recently, graph neural networks (GNN) techniques have been widely utilized to deal with non-Euclidean data, such as graphs or manifolds. Thus, it is essential to combine the advantages of the powerful learning capability of GNN models and multi-view data. In this paper, we aim to provide a comprehensive survey of recent research works on GNN-based multi-view learning. In detail, we first provide a taxonomy of GNN-based multi-view learning methods according to the input form of models: multi-relation, multi-attribute and mixed. Then, we introduce the applications of multi-view learning, including recommendation systems, computer vision and so on. Moreover, several public datasets and open-source codes are introduced for implementation. Finally, we analyze the challenges of applying GNN models on various multi-view learning tasks and state new future directions in this field.

The increasing uptake of converter-interfaced generation (CIG) is changing power system dynamics, rendering them extremely dependent on fast and complex control systems. Regularly assessing the stability of these systems across a wide range of operating conditions is thus a critical task for ensuring secure operation. However, the simultaneous simulation of both fast and slow (electromechanical) phenomena, along with an increased number of critical operating conditions, pushes traditional dynamic security assessments (DSA) to their limits. While DSA has served its purpose well, it will not be tenable in future electricity systems with thousands of power electronic devices at different voltage levels on the grid. Therefore, reducing both human and computational efforts required for stability studies is more critical than ever. In response to these challenges, several advanced simulation techniques leveraging artificial intelligence (AI) have been proposed in recent years. AI techniques can handle the increased uncertainty and complexity of power systems by capturing the non-linear relationships between the system’s operational conditions and their stability without solving the set of algebraic-differential equations that model the system. Once these relationships are established, system stability can be promptly and accurately evaluated for a wide range of scenarios. While hundreds of research articles confirm that AI techniques are paving the way for fast stability assessments, many questions and issues must still be addressed, especially regarding the pertinence of studying specific types of stability with the existing AI-based methods and their application in real-world scenarios. In this context, this article presents a comprehensive review of AI-based techniques for stability assessments in power systems. Different AI technical implementations, such as learning algorithms and the generation and treatment of input data, are widely discussed and contextualized. Their practical applications, considering the type of stability, system under study, and type of applications, are also addressed. We review the ongoing research efforts and the AI-based techniques put forward thus far for DSA, contextualizing and interrelating them. We also discuss the advantages, limitations, challenges, and future trends of AI techniques for stability studies.