Article
Chemistry, Analytical
Keshav Thapa, Yousung Seo, Sung-Hyun Yang, Kyong Kim
Summary: The study focuses on classifying human activities and inferring human behavior using modern sensing technology. However, the issue of domain adaptation for inertial sensing-based human activity recognition is still challenging. The requirement of labeled training data for adapting classifiers to new individuals or devices is a significant barrier. We propose a semi-supervised HAR method that improves reconstruction and generation without changes to a pre-trained classifier, achieving competitive improvement in handling new and unlabeled activity.
Article
Engineering, Electrical & Electronic
Yousef Pourebrahim, Farbod Razzazi, Hossein Sameti
Summary: This paper proposes a semi-supervised method for speech emotion recognition based on autoencoders, which combines information from unlabeled samples and labeled samples using a maximum mean discrepancy cost function to reduce distribution differences. Experimental results show that the proposed method outperforms previous methods on different emotional speech datasets, with potential for wide applications.
DIGITAL SIGNAL PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Yuxun Qu, Yongqiang Tang, Xuebing Yang, Yanlong Wen, Wensheng Zhang
Summary: This paper proposes a novel context-aware mutual learning method for semi-supervised human activity recognition. It introduces a semi-supervised mutual learning framework to alleviate the overfitting problem and proposes a distribution-preserving loss to hinder distribution deviation. It also adopts contextual information through a context-aware aggregation module. Experimental results show that the proposed method outperforms four typical methods in semi-supervised human activity recognition.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Chemistry, Analytical
Seungmin Oh, Akm Ashiquzzaman, Dongsu Lee, Yeonggwang Kim, Jinsul Kim
Summary: In recent years, deep learning models have been used for research in human activity recognition, but the lack of labeled data has led to slow development. Existing methods heavily rely on manual data collection and labeling, resulting in slow and biased processes. By proposing a solution using semi-supervised active transfer learning to reduce labeling tasks, performance was improved while reducing the amount of labeling required.
Article
Computer Science, Artificial Intelligence
Xiao Wang, Daisuke Kihara, Jiebo Luo, Guo-Jun Qi
Summary: The study introduces a new EnAET framework to enhance semi-supervised learning methods with self-supervised information. Experimental results demonstrate that the EnAET framework significantly improves the performance of semi-supervised algorithms, even in scenarios with a limited number of images, and can greatly enhance supervised learning as well.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Engineering, Aerospace
Zhifei Xi, Yue Lyu, Yingxin Kou, Zhanwu Li, You Li
Summary: This paper proposes an air combat target maneuver recognition model based on an online ensemble semi-supervised classification framework. The model achieves target maneuver recognition through offline training of basic classifiers, initialization of online recognition model, online recognition of target maneuver, and online model updates. Experimental results demonstrate that the proposed method achieves higher classification accuracy compared to other semi-supervised models and supervised models.
CHINESE JOURNAL OF AERONAUTICS
(2023)
Article
Engineering, Electrical & Electronic
Sungtae An, Asim H. Gazi, Omer T. Inan
Summary: In this article, a novel framework called DynaLAP is proposed for activity recognition in fixed protocols using a semi-supervised variational recurrent neural network (VRNN). Experimental results on two datasets demonstrate that DynaLAP outperforms previous methods in terms of performance.
IEEE SENSORS JOURNAL
(2022)
Article
Computer Science, Artificial Intelligence
Esra Adiyeke, Mustafa Gokce Baydogan
Summary: This paper introduces alternative semi-supervised tree-based strategies that are robust to scale differences both in terms of feature and target variables. Proposing the use of a scale-invariant proximity measure by means of tree-based ensembles to preserve the original characteristics of the data, the paper updates the classical tree derivation procedure to a multi-criteria form to resolve scale inconsistencies.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Interdisciplinary Applications
Spyridon Plakias, Yiannis S. Boutalis
Summary: This study proposes a novel information processing Deep Learning framework for un-supervised fault detection applications, using only samples from the normal class during the training process. The framework is based on an ensemble of Auto-Encoders, and the final decision for the normality of testing samples is made through a soft voting process, considering the confidence of each Auto-Encoder. The effectiveness of the proposed model is demonstrated through simulation results with widely used fault detection datasets.
COMPUTERS IN INDUSTRY
(2022)
Article
Computer Science, Information Systems
Juan Pablo Consuegra-Ayala, Yoan Gutierrez, Yudivian Almeida-Cruz, Manuel Palomar
Summary: This paper presents a two-phase optimization system that utilizes Auto-ML tools to solve classification problems and generate more robust classifiers. The experimental results show that ensembling a subset of already tested models can build a better solution, and ensuring diversity using the double-fault measure produces better results.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Wei Wang, Xiaoyang Suo, Xiangyu Wei, Bin Wang, Hao Wang, Hong-Ning Dai, Xiangliang Zhang
Summary: Graph Auto-Encoder is a framework for unsupervised learning on graph-structured data. However, it is not applicable for heterogeneous graphs that contain more abundant semantic information. Therefore, this work proposes a novel HGATE method for unsupervised representation learning on heterogeneous graph-structured data.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Information Systems
Riccardo Presotto, Gabriele Civitarese, Claudio Bettini
Summary: Federated Learning (FL) is a promising paradigm for sensor-based Human Activity Recognition (HAR) to address privacy and scalability issues. However, non-IID data and personalization pose challenges in the HAR domain. This work proposes SS-FedCLAR, a novel framework combining Federated Clustering and Semi-Supervised learning to enhance recognition rates.
PERVASIVE AND MOBILE COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Jovan Chavoshinejad, Seyed Amjad Seyedi, Fardin Akhlaghian Tab, Navid Salahian
Summary: Semi-supervised nonnegative matrix factorization combines the strengths of matrix factorization in learning part-based representation and can achieve high learning performance with limited labeled data and a large amount of unlabeled data. Recent research focuses on utilizing self-supervised learning to enhance semi-supervised learning. This paper proposes an effective Self-Supervised Semi-Supervised Nonnegative Matrix Factorization (S4NMF) model that directly extracts a consensus result from ensembled NMFs with similarity and dissimilarity regularizations. Experimental results on standard benchmark datasets demonstrate the effectiveness of the proposed model in semi-supervised clustering.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Information Systems
Xufeng Niu, Wenping Ma
Summary: To tackle the challenging task of high-dimensional data classification with limited labeled samples, we propose two semi-supervised learning models, SSRS and its adaptive version, ASSRS. These models address the unique characteristics of high-dimensional data by selecting subspaces of sample and feature dimensions and reducing dimensions. By incorporating sample-labeling auxiliary algorithm, adaptive sample subspace algorithm, and adaptive weight voting rule, ASSRS outperforms SSRS in terms of performance. Experiments demonstrate that SSRS and ASSRS perform better than other competitive algorithms and accurately label samples in datasets with limited labeled samples.
INFORMATION SCIENCES
(2023)
Article
Ecology
Benjamin Rowe, Philip Eichinski, Jinglan Zhang, Paul Roe
Summary: This paper introduces a technique for learning a general feature representation from unlabelled audio using auto-encoders, and finds marginal improvements over traditional "acoustic indices" at a 1-second timescale.
ECOLOGICAL INFORMATICS
(2021)
Article
Biochemical Research Methods
Robson P. Bonidia, Douglas S. Domingues, Danilo S. Sanches, Andre C. P. L. F. de Carvalho
Summary: This paper presents a new package called MathFeature, which is capable of extracting relevant numerical information from biological sequences. The features extracted by MathFeature showed high performance and robustness. It also provides descriptors not available in other packages and allows non-experts to use feature extraction techniques.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Computer Science, Information Systems
Gabriel J. Aguiar, Everton J. Santana, Andre C. P. F. L. de Carvalho, Sylvio Barbon Junior
Summary: The study examines the recommendation of method/base learner for multiple outputs and demonstrates the performance of meta-model through meta-learning experiments. The meta-models recommended high predictive performance solutions for multi-target regression tasks, including recommendations for real-world applications.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Adriano Rivolli, Luis P. F. Garcia, Carlos Soares, Joaquin Vanschoren, Andre C. P. L. F. de Carvalho
Summary: Meta-learning is increasingly utilized for recommending machine learning algorithms and configurations, but the inconsistency in describing and computing meta-features hampers reproducibility and comparability of empirical studies. This paper addresses this issue by systematizing and standardizing data characterization measures for classification datasets in meta-learning, while also providing an extensive list of meta-features and characterization tools as a guide for practitioners. The survey highlights the particularities and potential future directions in the development of meta-features for meta-learning.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Biochemical Research Methods
Robson P. Bonidia, Anderson P. Avila Santos, Breno L. S. de Almeida, Peter F. Stadler, Ulisses N. da Rocha, Danilo S. Sanches, Andre C. P. L. F. de Carvalho
Summary: Recent technological advances have led to exponential growth of biological sequence data and extraction of meaningful information through Machine Learning algorithms. This has improved our understanding of fatal diseases and led to the development of innovative solutions. However, ML-based approaches to biological data require representative features, and the manual processes of feature engineering and algorithm selection can be time-consuming. To tackle this, BioAutoML is introduced as an automated ML pipeline for biological data, with automated feature engineering and algorithm recommendation, saving time and improving predictive performance.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Materials Science, Multidisciplinary
Saulo Martiello Mastelini, Daniel R. Cassar, Edesio Alcobaca, Tiago Botari, Andre C. P. L. F. de Carvalho, Edgar D. Zanotto
Summary: Chalcogenide glasses are widely used in microelectronic and photonic devices due to their unique optical and electronic functionalities. Understanding the composition-property relationships is crucial for expanding their range of compositions and applications. By collecting a large quantity of data and utilizing machine learning algorithms, predictive models were induced to reveal the significant impact of key elements on the properties of chalcogenide glasses. These models can be utilized to design novel chalcogenide glasses with desired combinations of properties.
Article
Physics, Multidisciplinary
Robson P. Bonidia, Anderson P. Avila Santos, Breno L. S. de Almeida, Peter F. Stadler, Ulisses Nunes da Rocha, Danilo S. Sanches, Andre C. P. L. F. de Carvalho
Summary: In recent years, the exponential growth in sequencing projects has posed new challenges for biological sequence analysis. Machine learning algorithms have been explored for analyzing and classifying biological sequences, despite the difficulty in finding suitable methods. This study proposes a novel Tsallis entropy-based feature extractor for classifying biological sequences, which has been proven to be effective and robust in terms of generalization through five case studies.
Article
Computer Science, Artificial Intelligence
Tomas Horvath, Rafael G. Mantovani, Andre C. P. L. F. de Carvalho
Summary: Meta-learning, a concept from automated machine learning, recommends suitable settings for given datasets based on the assumption that optimal settings for one dataset are also suitable for similar datasets. This paper introduces a new perspective on using PCA meta-features, which are rarely used despite their good descriptive characteristics and easy computation. A novel meta-learning approach using the DTW similarity measure is proposed for computing dataset similarities based on cumulative variances explained by principal components. Experimental results on 50 real-world datasets demonstrate the potential of combining PCA and DTW in meta-learning and encourage further investigation.
APPLIED SOFT COMPUTING
(2023)
Review
Computer Science, Artificial Intelligence
Angelo G. Menezes, Gustavo de Moura, Cezanne Alves, Andre C. P. L. F. de Carvalho
Summary: Continual learning focuses on learning consecutive tasks without losing performance on previously learned tasks. While most research has been on incremental classification tasks, continual object detection deserves more attention due to its wide range of applications. It is more complex than traditional classification, with instances of unknown classes appearing in subsequent tasks, resulting in missing annotations and conflicts with background labels.
Article
Genetics & Heredity
Muhammad Kabiru Nata'ala, Anderson P. Avila Santos, Jonas Coelho Kasmanas, Alexander Bartholomaeus, Joao Pedro Saraiva, Sandra Godinho Silva, Tina Keller-Costa, Rodrigo Costa, Newton C. M. Gomes, Andre Carlos Ponce de Leon Ferreira de Carvalho, Peter F. Stadler, Danilo Sipoli Sanches, Ulisses Nunes da Rocha
Summary: The MarineMetagenomeDB is a valuable resource for researchers to find marine metagenome samples with curated metadata and facilitate meta-studies involving marine microbiomes. The database contains over 11,000 marine metagenomes from all oceans and provides user-friendly search and download tools.
ENVIRONMENTAL MICROBIOME
(2022)
Article
Computer Science, Information Systems
Gean T. Pereira, Iury B. A. Santos, Luis P. F. Garcia, Thierry Urruty, Muriel Visani, Andre C. P. L. F. de Carvalho
Summary: This paper proposes a Prediction-based and interpretable Meta-Learning method called MbML-NAS, which can generalize to different search spaces and datasets using less data. The method uses interpretable meta-features extracted from neural architectures and regression models as meta-predictors to infer Convolutional Networks performances.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Saulo Martiello Mastelini, Felipe Kenji Nakano, Celine Vens, Andre Carlos Ponce de Leon Ferreira de Carvalho
Summary: As the amount of data continues to grow, traditional machine learning algorithms are no longer capable of handling large volumes of data. This paper proposes a new online machine learning algorithm called online extra trees, which is based on decision tree ensembles. The algorithm combines subbagging, random tree split points, and model trees to achieve high accuracy while reducing computational costs.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Linguistics
Hidelberg O. Albuquerque, Ellen Souza, Carlos Gomes, Matheus Henrique de C. Pinto, Ricardo P. S. Filho, Rosimeire Costa, Vinicius Teixeira de M. Lopes, Nadia F. F. da Silva, Andre C. P. L. F. de Carvalho, Adriano L. I. Oliveira
Summary: Named Entity Recognition (NER) is a crucial task in Natural Language Processing, with various applications and research interest. However, resources for languages like Portuguese are lacking. This study aims to map NER techniques and resources for Portuguese. A total of 447 primary studies were retrieved, and 45 were included in the review. Comparative analysis, new corpora, and commonly used techniques and algorithms were identified.
PROCESAMIENTO DEL LENGUAJE NATURAL
(2023)
Review
Dentistry, Oral Surgery & Medicine
Anna Luiza Damaceno Araujo, Viviane Mariano da Silva, Maira Suzuka Kudo, Eduardo Santos Carlos de Souza, Cristina Saldivia-Siracusa, Daniela Giraldo-Roldan, Marcio Ajudarte Lopes, Pablo Agustin Vargas, Syed Ali Khurram, Alexander T. Pearson, Luiz Paulo Kowalski, Andre Carlos Ponce de Leon Ferreira de Carvalho, Alan Roger Santos-Silva, Matheus Cardoso Moraes
Summary: This paper aims to provide oral pathologists, oral medicinists, and head and neck surgeons with a theoretical and conceptual foundation of artificial intelligence-based diagnostic approaches, with a special focus on convolutional neural networks, the state-of-the-art in artificial intelligence and deep learning. Artificial intelligence models and networks can learn and process dense information in a short time, leading to an efficient, objective, and accurate clinical and histopathological analysis, which can be useful to improve treatment modalities and prognostic outcomes.
JOURNAL OF ORAL PATHOLOGY & MEDICINE
(2023)
Article
Microbiology
Anderson Paulo Avila Santos, Muhammad Kabiru Nata'ala, Jonas Coelho Kasmanas, Alexander Bartholomaeus, Tina Keller-Costa, Stephanie D. Jurburg, Tamara Tal, Amelia Camarinha-Silva, Joao Pedro Saraiva, Andre Carlos Ponce de Leon Ferreira de Carvalho, Peter F. Stadler, Danilo Sipoli Sanches, Ulisses Rocha
Summary: The Animal-Associated Metagenome Metadata Database (AAMDB) facilitates the identification and reuse of publicly available non-human, animal-associated metagenomic data and metadata. The database contains 10,885 metagenomes associated with 165 different species from 65 countries. The majority of the metagenomes are associated with animals used for medical research or human consumption. The database provides valuable information on the biogeography and diversity of animal-associated metagenomes.
Article
Computer Science, Artificial Intelligence
Saulo Martiello Mastelini, Bruno Veloso, Max Halford, Andre Carlos Ponce de Leon Ferreira de Carvalho, Joao Gama
Summary: This paper proposes a graph-based online search index algorithm called SWINN for speeding up nearest neighbor search (NNS) in potentially never-ending and dynamic data stream tasks. SWINN significantly outperforms a naive complete scan of the data buffer while maintaining competitive search recall values. It also proves to be effective against popular online machine learning algorithms when applied to online classification and regression tasks.
INFORMATION FUSION
(2024)
Article
Computer Science, Artificial Intelligence
Rui Lv, Dingheng Wang, Jiangbin Zheng, Zhao-Xu Yang
Summary: In this paper, the authors investigate tensor decomposition for neural network compression. They analyze the convergence and precision of tensor mapping theory, validate the rationality of tensor mapping and its superiority over traditional tensor approximation based on the Lottery Ticket Hypothesis. They propose an efficient method called 3D-KCPNet to compress 3D convolutional neural networks using the Kronecker canonical polyadic (KCP) tensor decomposition. Experimental results show that 3D-KCPNet achieves higher accuracy compared to the original baseline model and the corresponding tensor approximation model.
Article
Computer Science, Artificial Intelligence
Xiangkun He, Zhongxu Hu, Haohan Yang, Chen Lv
Summary: In this paper, a novel constrained multi-objective reinforcement learning algorithm is proposed for personalized end-to-end robotic control with continuous actions. The approach trains a single model using constraint design and a comprehensive index to achieve optimal policies based on user-specified preferences.
Article
Computer Science, Artificial Intelligence
Zhijian Zhuo, Bilian Chen, Shenbao Yu, Langcai Cao
Summary: In this paper, a novel method called Expansion with Contraction Method for Overlapping Community Detection (ECOCD) is proposed, which utilizes non-negative matrix factorization to obtain disjoint communities and applies expansion and contraction processes to adjust the degree of overlap. ECOCD is applicable to various networks with different properties and achieves high-quality overlapping community detection.
Article
Computer Science, Artificial Intelligence
Yizhe Zhu, Chunhui Zhang, Jialin Gao, Xin Sun, Zihan Rui, Xi Zhou
Summary: In this work, the authors propose a Contrastive Spatio-Temporal Distilling (CSTD) approach to improve the detection of high-compressed deepfake videos. The approach leverages spatial-frequency cues and temporal-contrastive alignment to fully exploit spatiotemporal inconsistency information.
Review
Computer Science, Artificial Intelligence
Laijin Meng, Xinghao Jiang, Tanfeng Sun
Summary: This paper provides a review of coverless steganographic algorithms, including the development process, known contributions, and general issues in image and video algorithms. It also discusses the security of coverless steganography from theoretical analysis to actual investigation for the first time.
Article
Computer Science, Artificial Intelligence
Yajie Bao, Tianwei Xing, Xun Chen
Summary: Visual question answering requires processing multi-modal information and effective reasoning. Neural-symbolic learning is a promising method, but current approaches lack uncertainty handling and can only provide a single answer. To address this, we propose a confidence based neural-symbolic approach that evaluates NN inferences and conducts reasoning based on confidence.
Article
Computer Science, Artificial Intelligence
Anh H. Vo, Bao T. Nguyen
Summary: Interior style classification is an interesting problem with potential applications in both commercial and academic domains. This project proposes a method named ISC-DeIT, which combines data-efficient image transformer architectures and knowledge distillation, to address the interior style classification problem. Experimental results demonstrate a significant improvement in predictive accuracy compared to other state-of-the-art methods.
Article
Computer Science, Artificial Intelligence
Shashank Kotyan, Danilo Vasconcellos Vargas
Summary: This article introduces a novel augmentation technique called Dynamic Scanning Augmentation to improve the accuracy and robustness of Vision Transformer (ViT). The technique leverages dynamic input sequences to adaptively focus on different patches, resulting in significant changes in ViT's attention mechanism. Experimental results demonstrate that Dynamic Scanning Augmentation outperforms ViT in terms of both robustness to adversarial attacks and accuracy against natural images.
Article
Computer Science, Artificial Intelligence
Hiba Alqasir, Damien Muselet, Christophe Ducottet
Summary: The article proposes a solution to improve the learning process of a classification network by providing shape priors, reducing the need for annotated data. The solution is tested on cross-domain digit classification tasks and a video surveillance application.
Article
Computer Science, Artificial Intelligence
Dexiu Ma, Mei Liu, Mingsheng Shang
Summary: This paper proposes a method using neural dynamics solvers to solve infinity-norm optimization problems. Two improved solvers are constructed and their effectiveness and superiority are demonstrated through theoretical analysis and simulation experiments.
Article
Computer Science, Artificial Intelligence
Francesco Gregoretti, Giovanni Pezzulo, Domenico Maisto
Summary: Active Inference is a computational framework that uses probabilistic inference and variational free energy minimization to describe perception, planning, and action. cpp-AIF is a header-only C++ library that provides a powerful tool for implementing Active Inference for Partially Observable Markov Decision Processes through multi-core computing. It is cross-platform and improves performance, memory management, and usability compared to existing software.
Article
Computer Science, Artificial Intelligence
Zelin Ying, Dawei Cheng, Cen Chen, Xiang Li, Peng Zhu, Yifeng Luo, Yuqi Liang
Summary: This paper proposes a novel stock market trends prediction framework called SMART, which includes a self-supervised stock technical data sequence embedding model S3E. By training with multiple self-supervised auxiliary tasks, the model encodes stock technical data sequences into embeddings and uses the learned sequence embeddings for predicting stock market trends. Extensive experiments on China A-Shares market and NASDAQ market prove the high effectiveness of our model in stock market trends prediction, and its effectiveness is further validated in real-world applications in a leading financial service provider in China.
Article
Computer Science, Artificial Intelligence
Hao Li, Hao Jiang, Dongsheng Ye, Qiang Wang, Liang Du, Yuanyuan Zeng, Liu Yuan, Yingxue Wang, C. Chen
Summary: DHGAT1, a dynamic hyperbolic graph attention network, utilizes hyperbolic metric properties to embed dynamic graphs. It employs a spatiotemporal self-attention mechanism and weighted node representations, resulting in excellent performance in link prediction tasks.
Article
Computer Science, Artificial Intelligence
Jiehui Huang, Zhenchao Tang, Xuedong He, Jun Zhou, Defeng Zhou, Calvin Yu-Chian Chen
Summary: This study proposes a progressive learning multi-scale feature blending model for image deraining tasks. The model utilizes detail dilation and texture extraction to improve the restoration of rainy images. Experimental results show that the model achieves near state-of-the-art performance in rain removal tasks and exhibits better rain removal realism.
Article
Computer Science, Artificial Intelligence
Lizhi Liu, Zilin Gao, Yinhe Wang, Yongfu Li
Summary: This paper proposes a novel discrete-time interconnected model for depicting complex dynamical networks. The model consists of nodes and edges subsystems, which consider the dynamic characteristic of both nodes and edges. By designing control strategies and coupling modes, the stabilization and synchronization of the network are achieved. Simulation results demonstrate the effectiveness of the proposed methods.