Related references
Note: Only part of the references are listed.
Article
Computer Science, Artificial Intelligence
Wentao Cui et al.
Summary: Contrastive learning, as a self-supervised method, has achieved significant success by learning latent semantic class information. However, treating instances as classes hinders the learning of true latent semantic classes due to the presence of hard negatives. To address this, we propose a new contrastive learning framework called TNCC, which uses a loss based on Student-t distribution and incorporates a neighbor consistency constraint. Experimental results on benchmark datasets demonstrate the effectiveness of the proposed framework.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Hamit Taner Unal et al.
Summary: We introduce a new neural architecture called Neural Logic Circuits (NLC), which is an evolutionary, weightless, and learnable model inspired by the brain's neuroplasticity. This model achieves learning through the evolution of its architecture by reorganizing synaptic connections and generating artificial neurons as logic gates. Unlike Artificial Neural Networks (ANN), our model achieves generalization ability without intensive weight training and focuses computational resources on building optimal network architecture. Experimental results show the remarkable superiority of our initial model, NLCv1, on well-known binary classification datasets compared to modern and competitive machine learning algorithms.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Qian Wang et al.
Summary: By utilizing unsupervised contrastive learning with prompt mechanism, the proposed CLSEP provides effective sentence embedding representation, and the novel data augmentation strategy PWVA retains more semantic information. The effectiveness of the proposed methods is verified on the NLI task.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Lei Jiang et al.
Summary: As a popular research field of sentiment analysis, aspect-based sentiment analysis aims to analyze emotional expressions in different aspects. However, current research lacks precision in dividing aspects, leading to semantic overlap issues. To address these problems, we propose the concept of non-dependent aspects and a method for their division based on analyzing aspect dependencies. Theoretical analysis and real-world experiments demonstrate that sentiment analysis based on non-dependent aspects provides more accurate results compared to traditional methods.
KNOWLEDGE-BASED SYSTEMS
(2023)
Review
Computer Science, Artificial Intelligence
Zheng Wang et al.
Summary: With the increasing popularity of social websites and mobile applications, online videos shared by customers providing their thoughts and reviews on products have become more prevalent. This paper proposes a novel method for analyzing customer sentiment from these online videos. The method involves applying latent Dirichlet allocation (LDA) modeling for topic identification and using a newly designed multi-attention bi-directional LSTM (BLSTM(MA)) to determine sentiment polarity. The research is of great practical value for company managers and researchers to better understand customer opinions on specific products.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Ioannis Almalis et al.
Summary: This paper introduces a deep learning solution for sentiment analysis that performs well on previously unseen tweet data. Additionally, the study provides a novel research on predicting specific economic sectors affected by news articles.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Sangjin Park et al.
Summary: This study provides two approaches, a deep learning model based on the LSTM network architecture and an interpretable machine learning model, for better economic prediction and decision-making. The research found that private debt in developed economies and government debt in emerging economies are major factors limiting future economic growth. In terms of the economic impact of COVID-19, reduced interest rates and increased government debt have increased the probability of a crisis in some emerging economies in the future.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Jiyang Zhang et al.
Summary: This study proposes a novel feature-learning-based method called CA-SupCon for addressing the class-imbalance issue in fault diagnosis. By utilizing supervised contrastive learning and a class-aware sampler, this method not only enlarges the feature distance between minority fault conditions but also obtains a more discriminative feature space.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Isidoros Perikos et al.
Summary: The article introduces a novel, interpretable HMM-based method for recognizing sentiments in text, which has been tested under various architectures, training methods, orders, and ensembles, showing competitive performance and outperforming traditional HMM methods.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Kehui Song et al.
Summary: This paper introduces an interpretable medical decision support system that integrates historical cases and expert opinions to establish a medical knowledge base for providing relevant recommendations. The system outperforms others in terms of specificity, sensitivity, and F-1 score, contributing to reliable support for diagnosis prediction.
KNOWLEDGE-BASED SYSTEMS
(2021)
Review
Engineering, Electrical & Electronic
Wojciech Samek et al.
Summary: With the increasing demand for explainable artificial intelligence (XAI) due to the successful usage of machine learning, particularly deep neural networks, this work aims to provide an overview of the field, test interpretability algorithms, and demonstrate successful usage in application scenarios.
PROCEEDINGS OF THE IEEE
(2021)
Article
Computer Science, Artificial Intelligence
Daniel Zeltner et al.
Summary: The research explores a new approach to address the interpretability issue of deep neural networks by combining soft continuous logic and multi-criteria decision-making tools to improve transparency by reducing the black-box nature of neural models. Experimental results demonstrate that Squashing functions perform well in neural networks, similar to conventional activation functions, and achieve high performance in solving simple classification tasks.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Jiyao Wei et al.
Article
Computer Science, Artificial Intelligence
Orsolya Csiszar et al.
KNOWLEDGE-BASED SYSTEMS
(2020)
Article
Computer Science, Artificial Intelligence
Cynthia Rudin
NATURE MACHINE INTELLIGENCE
(2019)
Article
Computer Science, Artificial Intelligence
Wojciech Samek et al.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2017)
Article
Computer Science, Artificial Intelligence
Kumar Ravi et al.
KNOWLEDGE-BASED SYSTEMS
(2015)
Article
Biochemical Research Methods
Trias Thireou et al.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2007)
