Article
Computer Science, Information Systems
Fatima-zahra El-Alami, Said Ouatik El Alaoui, Noureddine En Nahnahi
Summary: This paper investigates the potential of pre-trained Arabic BERT model for learning contextual sentence representations and its application in Arabic text multi-class categorization. Experimental results show that fine-tuned AraBERT model achieves state-of-the-art performance.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Zhen Wang, Liu Liu, Yiqun Duan, Dacheng Tao
Summary: This study proposes the task of few-shot streaming label learning (FSLL) and introduces a meta-learning framework (SIN) to adapt to this task. SIN leverages label semantic representation to regularize the output space and acquires labelwise meta-knowledge through meta-learning. Additionally, SIN incorporates a label decision module and meta-threshold loss function to determine the optimal confidence thresholds for each new label. Experimental results demonstrate that SIN outperforms the prior state-of-the-art methods on FSLL.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Information Systems
Xiaoyan Wang, Hongmei Wang, Daming Zhou
Summary: Research on few-shot learning aims to learn new concepts from a small number of labeled samples. A novel feature transformation network (FTN) is proposed for few-shot image classification, which introduces attention-based affinity matrix to enhance sample representation focusing on target attributes.
Article
Engineering, Electrical & Electronic
Wen Jiang, Kai Huang, Jie Geng, Xinyang Deng
Summary: This paper proposes a novel few-shot learning method called multi-scale metric learning (MSML) to tackle the classification problem in few-shot learning by extracting multi-scale features and learning multi-scale relationships. The method introduces a feature pyramid structure and a multi-scale relation generation network, and optimizes the deep network with the intra-class and inter-class relation loss, achieving superior performance in experimental results on mini ImageNet and tiered ImageNet.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Haigang Zhang, Xianglong Meng, Weipeng Cao, Ye Liu, Zhong Ming, Jinfeng Yang
Summary: Multi-label Zero-shot Learning (ZSL) is a more reasonable and realistic approach than standard single-label ZSL as it considers the coexistence of multiple objects in real-life images. Intra-class feature entanglement affects the alignment of visual and semantic features, making it difficult for the model to recognize unseen samples comprehensively. Existing multi-label ZSL methods focus on attention-based refinement and decoupling of visual features, but overlook the relationship between label semantics. This paper proposes a method that utilizes label correlations and builds a weighted semantic graph to guide visual feature extraction, achieving improved performance compared to state-of-the-art models.
Article
Computer Science, Artificial Intelligence
Tianshui Chen, Liang Lin, Riquan Chen, Xiaolu Hui, Hefeng Wu
Summary: This research introduces a knowledge-guided graph routing (KGGR) framework that combines prior knowledge of statistical label correlations with deep neural networks to address challenges in multi-label recognition tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Ziming Liu, Song Guo, Jingcai Guo, Yuanyuan Xu, Fushuo Huo
Summary: Multi-label zero-shot learning aims at recognizing multiple unseen labels for each input sample, but existing methods often neglect minor classes and result in inadequate attention. This paper proposes a novel unbiased framework that balances the training process by considering class-specific regions and strengthens the correlation among semantic representations.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Chemistry, Analytical
Depei Wang, Zhuowei Wang, Lianglun Cheng, Weiwen Zhang
Summary: In this paper, the authors propose SumFS, a meta-learning framework for text classification that utilizes extractive summarization and improved local vocabulary features to achieve domain adaptation with limited labeled data. Experimental results show that SumFS can reduce input features while maintaining or improving accuracy, and significantly decrease training time.
Article
Chemistry, Multidisciplinary
Zhe Ren, Xizhong Qin, Wensheng Ran
Summary: In this paper, an enhanced Span and Label semantic representation method is proposed for Chinese few-shot Named Entity Recognition (SLNER). The method uses two encoders, one to encode the text and its spans to obtain enhanced span representations, and the other to encode the label names to obtain label representations. The model learns to match span representations with label representations, and experimental results show that it outperforms previous state-of-the-art methods in few-shot settings.
APPLIED SCIENCES-BASEL
(2023)
Article
Acoustics
Huang Xie, Tuomas Virtanen
Summary: This paper investigates zero-shot learning in audio classification using semantic embeddings extracted from textual labels and sentence descriptions, demonstrating the effectiveness of a bilinear compatibility framework and deep acoustic embeddings in improving classification performance. By involving semantically close sound classes in training and concatenating label/sentence embeddings from different language models, the results are further enhanced.
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Gong Cheng, Chunbo Lang, Junwei Han
Summary: Conventional deep CNN-based segmentation approaches are difficult to generalize to unseen categories, so few-shot segmentation is developed to handle this issue. However, existing methods often overfit base categories and produce inaccurate segmentation boundaries. In this paper, a Holistic Prototype Activation (HPA) network is proposed to alleviate these problems by introducing novel designs, such as a training-free scheme, a Prototype Activation Module (PAM), and a Cross-Referenced Decoder (CRD). Experimental results on standard few-shot segmentation benchmarks and extended tasks demonstrate the effectiveness, flexibility, and versatility of the proposed method. The code is publicly available.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Alami Hamza, Said El Alaoui Ouatik, Khalid Alaoui Zidani, Noureddine En-Nahnahi
Summary: This paper presents a duplicate question detection method based on contextual word representation, question classification, and self-attention, achieving good performance in experiments.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Engineering, Civil
Xianda Xu, Xing Xu, Fumin Shen, Yujie Li
Summary: Autonomous driving relies on accurate visual recognition of surrounding objects, and few-shot image classification is employed to recognize rarely seen objects. This paper introduces a Semantic-Aligned Attention (SAA) mechanism to refine feature embedding in existing embedding and metric-learning approaches, showing competitive improvements in both few-shot and zero-shot classification tasks on benchmark datasets.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Yuhan Chai, Lei Du, Jing Qiu, Lihua Yin, Zhihong Tian
Summary: The continuous increase and spread of malware have caused immeasurable losses to social enterprises and even the country, especially unknown malware. Most existing methods use predefined class samples to train models, which cannot handle unknown malware detection. In this paper, we propose a Dynamic Prototype Network based on Sample Adaptation for few-shot malware detection (DPNSA) to address this issue. Our method outperforms the existing models and achieves significant improvement in malware detection.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Boyu Yang, Fang Wan, Chang Liu, Bohao Li, Xiangyang Ji, Qixiang Ye
Summary: This article introduces a part-based semantic transform (PST) method to align object semantics in support and query images through semantic decomposition and matching, effectively addressing the semantic misalignment issue in few-shot semantic segmentation.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Ankita Gandhi, Kinjal Adhvaryu, Soujanya Poria, Erik Cambria, Amir Hussain
Summary: This survey paper explores the importance and recent advancements in sentiment analysis and multimodal sentiment analysis in the fields of artificial intelligence and natural language processing. It compares various fusion architectures in terms of MSA categories and presents interdisciplinary applications and future research directions.
INFORMATION FUSION
(2023)
Article
Computer Science, Artificial Intelligence
Tian-Hui You, Ling-Ling Tao, Erik Cambria
Summary: This study proposes a hotel ranking model based on online textual reviews, considering the differences in the number of reviews on different aspects. The model utilizes sentiment analysis to assist tourists in making desirable decisions on hotel selection.
INTERNATIONAL JOURNAL OF INFORMATION TECHNOLOGY & DECISION MAKING
(2023)
Article
Computer Science, Artificial Intelligence
Javier Torregrosa, Sergio D'Antonio-Maceiras, Guillermo Villar-Rodriguez, Amir Hussain, Erik Cambria, David Camacho
Summary: Political tensions have increased in Europe since the beginning of the new century, leading to social movements and political changes in various countries. This study examines the political discourse and underlying tensions during Madrid's elections in May 2021, using a mixed methodology approach. The findings suggest that the electoral campaign is not as negative as perceived by the citizens, and that ideologically extreme parties tend to use more aggressive language.
COGNITIVE COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Wei Sun, Shaoxiong Ji, Erik Cambria, Pekka Marttinen
Summary: Human coders assign standardized medical codes to clinical documents, but it is prone to errors and requires significant effort. Automated medical coding methods using machine learning, such as deep neural networks, have been developed. However, challenges still exist due to code association complexity, noise in lengthy documents, and imbalanced class problem. In this study, we propose a novel neural network model called the Multitask Balanced and Recalibrated Neural Network to address these issues. Experiments on a real-world clinical dataset called MIMIC-III demonstrate that our model outperforms competitive baselines.
ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Xiaoshi Zhong, Erik Cambria
Summary: Time information is crucial in the fields of data mining, information retrieval, and natural language processing. Time expression recognition and normalization (TERN) serves as a fundamental task for other linguistic tasks. This survey reviews previous research, provides an overview of time expression analysis development, and explores the role of time expressions in different domains. Three methods for TERN development are discussed: rule-based, traditional machine-learning, and deep-learning. Additionally, useful datasets, software, and potential future research directions are outlined.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Xulang Zhang, Rui Mao, Erik Cambria
Summary: Computational syntactic processing is a fundamental technique in natural language processing that transforms natural language into structured texts with syntactic features. This work surveys low-level syntactic processing techniques such as normalization, sentence boundary disambiguation, part-of-speech tagging, text chunking, and lemmatization, categorizes widely used methods, investigates challenges, and proposes future research directions.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Jingfeng Cui, Zhaoxia Wang, Seng-Beng Ho, Erik Cambria
Summary: Sentiment analysis, a research hotspot in natural language processing, has attracted significant attention and resulted in a growing number of research papers. Despite numerous literature reviews on sentiment analysis, there has been no dedicated survey examining the evolution of research methods and topics. This study fills this gap by conducting a comprehensive survey that combines keyword co-occurrence analysis and community detection algorithm. The survey compares and analyzes the connections between research methods and topics over the past two decades and uncovers hotspots and trends over time, providing valuable guidance for researchers. Furthermore, the paper offers practical insights, technical directions, limitations, and future research prospects in sentiment analysis.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Qika Lin, Rui Mao, Jun Liu, Fangzhi Xu, Erik Cambria
Summary: Knowledge graph completion (KGC) is crucial for many downstream applications. Existing language model-based methods for KGC often overlook the importance of modeling the deeper semantic information, such as topology contexts and logical rules. In this paper, we propose a unified framework FTL-LM that effectively incorporates topology contexts and logical rules in language models, and experimental results demonstrate its superiority over the state-of-the-art methods.
INFORMATION FUSION
(2023)
Article
Computer Science, Information Systems
Erik Cambria, Lorenzo Malandri, Fabio Mercorio, Mario Mezzanzanica, Navid Nobani
Summary: The field of explainable artificial intelligence (XAI) has gained increasing importance in recent years. However, existing research often overlooks the role of natural language in generating explanations. This survey reviews 70 XAI papers published between 2006 and 2021 and evaluates their readiness in terms of natural language explanations. The results show that only a few recent studies have considered using natural language for communication with end users or implemented methods for generating natural language explanations.
INFORMATION PROCESSING & MANAGEMENT
(2023)
Article
Computer Science, Information Systems
Kelvin Du, Frank Xing, Erik Cambria
Summary: Combining symbolic and subsymbolic methods has emerged as a promising strategy in tackling increasingly complex AI research tasks. This study presents a targeted aspect-based financial sentiment analysis hybrid model that incorporates multiple lexical knowledge sources into the fine-tuning process of pre-trained transformer models. Experimental results demonstrate that knowledge-enabled models systematically improve aspect sentiment analysis performance and even outperform state-of-the-art results.
ACM TRANSACTIONS ON MANAGEMENT INFORMATION SYSTEMS
(2023)
Article
Computer Science, Cybernetics
Luna Ansari, Shaoxiong Ji, Qian Chen, Erik Cambria
Summary: Changes in human lifestyle have led to an increase in depression cases. Automated detection methods are effective in identifying depressed individuals. Ensemble models outperform hybrid models for depression detection.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Ruicheng Liu, Rui Mao, Anh Tuan Luu, Erik Cambria
Summary: The task of resolving repeated objects in natural languages, known as coreference resolution, is an important part of modern natural language processing. It is classified into entity coreference resolution and event coreference resolution based on the resolved objects. Predicting coreference connections and identifying mentions/triggers are the major challenges in coreference resolution due to the difficulty of implicit relationships in natural language understanding. In this survey, we review the current employed evaluation metrics, datasets, and methods, investigating 10 widely used metrics, 18 datasets, and 4 main technical trends. We believe that this work provides a comprehensive roadmap for understanding the past and the future of coreference resolution.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Zhaoxia Wang, Zhenda Hu, Seng-Beng Ho, Erik Cambria, Ah-Hwee Tan
Summary: This paper proposes a new explainable fine-grained multi-class sentiment analysis method called MiMuSA, which mimics human language understanding processes. It builds multiple knowledge bases to support sentiment understanding and can identify fine-grained multi-class sentiments. Experimental results show that MiMuSA outperforms other existing multi-class sentiment analysis methods in terms of accuracy and F1-Score.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Zhaoxia Wang, Zhenda Hu, Fang Li, Seng-Beng Ho, Erik Cambria
Summary: This research proposes a 3-phase hybrid model that utilizes both technical indicators and social media text sentiments as influence factors for stock trending prediction. The result shows that the proposed method has an accuracy of 73.41% and F1-score of 84.19%. The research not only demonstrates the merits of the proposed method, but also indicates that integrating social opinions with technical indicators is a right direction for enhancing the performance of learning-based stock market trending analysis methods.
COGNITIVE COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Deeksha Varshney, Asif Ekbal, Erik Cambria
Summary: This paper focuses on the neural-based interactive dialogue system that aims to engage and retain humans in long-lasting conversations. It proposes a new neural generative model that combines step-wise co-attention, self-attention-based transformer network, and an emotion classifier to control emotion and knowledge transfer during response generation. The results from quantitative, qualitative, and human evaluation show that the proposed models can generate natural and coherent sentences, capturing essential facts with significant improvement over emotional content.
KNOWLEDGE-BASED SYSTEMS
(2024)