Title: UNIFIED PREDICTIVE MODEL FOR PREDICTING THE CLOSING PRICE OF VARIOUS CRYPTOCURRENCIES
<br>Cover Date: 2025-06-01
<br>Cover Display Date: June 2025
<br>DOI: 10.24507/icicelb.16.06.589
<br>Description: This study introduces a unified predictive model for forecasting cryptocurrency closing prices, utilizing advanced machine learning techniques such as Support Vector Regression (SVR), Random Forest, and Long Short-Term Memory (LSTM) net-works. By integrating multiple cryptocurrency datasets through feature-level data fusion via concatenation, the model effectively captures the complex, nonlinear, and dynamic relationships characteristic of cryptocurrency markets. The experimental results reveal that the proposed model outperforms baseline models developed for individual cryptocurrencies, particularly with SVR. This enhanced performance is due to SVR’s ability to manage high-dimensional data and model intricate nonlinear patterns. While Random Forest and LSTM also demonstrate strong predictive capabilities, their effectiveness is more dependent on specific data characteristics and configurations. The integration of diverse data sources and the application of Min-Max normalization play a crucial role in enhancing prediction accuracy and model robustness. This approach allows the model to account for broader market dynamics, providing valuable insights for short-term and medium-term trading strategies and supporting informed decision-making for investors, traders, and analysts.
<br>Citations: 0
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: ENHANCING SENTIMENT CLASSIFICATION: A COMPARATIVE ANALYSIS OF SUPERVISED AND UNSUPERVISED METHODS FOR IMPROVING TRAINING DATA QUALITY
<br>Cover Date: 2025-05-01
<br>Cover Display Date: May 2025
<br>DOI: 10.24507/icicelb.16.05.471
<br>Description: This study evaluates the effectiveness of supervised and unsupervised methods in enhancing data quality for binary sentiment classification. Two datasets of hotel reviews from TripAdvisor were utilized: one for training polarity correction models and the other containing noisy labels for experimental evaluation. Supervised methods, including SVM with a linear kernel, Random Forest (RF), and Convolutional Neural Network (CNN), consistently outperformed unsupervised methods such as Standard K-means, K-means++, and Spherical K-means. Following the development of sentiment classifier models using the improved training set, SVM demonstrated the highest performance, achieving an accuracy and F1 score of 0.85, followed by RF and CNN. Among the unsupervised approaches, K-means++ yielded the best results, with an accuracy of 0.75 and an F1 score of 0.74. These findings highlight the superiority of supervised learning in sentiment classification tasks and underscore the critical importance of training set quality in enhancing model performance.
<br>Citations: 0
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: ENSEMBLE CLUSTERING METHOD FOR ASSEMBLING OF THAI DECIDED CIVIL CASES INTO SPECIFIC CLUSTERS
<br>Cover Date: 2025-03-01
<br>Cover Display Date: March 2025
<br>DOI: 10.24507/icicel.19.03.271
<br>Description: Civil cases often pertain to legal disputes between individuals or organizations. Following a judgment, civil cases are referred to as “decided cases” and the associated documents can be utilized for future legal determinations. One alternative method for managing these decided cases and making it easier to identify relevant decided cases that meet the user’s needs is to group relevant decided cases together. As a result, the purpose of this study was to offer an ensemble clustering method for finding and identifying the most relevant legal cases from a given collection that satisfy the needs of users. In our ensemble clustering, we employ well-known clustering methods such as k-means++, spherical k-means, and DBSCAN. Upon assessing the clustering quality measure (purity score), accuracy, and F1 score, the proposed method yielded good results. Furthermore, when comparing it to the baseline, the proposed method exhibits enhancements in the purity score, accuracy, and F1 score by 6.95%, 6.67%, and 6.95%, respectively.
<br>Citations: 0
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: A HYBRID METHOD OF ASPECT-BASED SENTIMENT ANALYSIS FOR HOTEL REVIEWS
<br>Cover Date: 2024-01-01
<br>Cover Display Date: January 2024
<br>DOI: 10.24507/icicel.18.01.59
<br>Description: The purpose of this study was to introduce a hybrid method of aspect-based sentiment analysis for hotel reviews. Hotel staff attentiveness, hotel cleanliness, value for money, and hotel location are all highly regarded hotel aspects. The proposed method is made up of two major components. BM25 is used in the first component to group the review sentences into the most relevant hotel aspect cluster. Word2Vec's skip-gram was utilized to generate the keywords relevant to each hotel aspect, which were then used as queries to organize review sentences into suitable hotel aspect cluster. Finally, hotel review sentences in each cluster are assigned a sentiment polarity as positive or negative using the sentiment polarity analyzer, which is an ensemble model comprised of five predictive models developed by C4.5 decision tree, Multinomial Naive Bayes (MNB), Support Vector Machines (SVM) with linear kernel, SVM with RBF kernel, and Logistic Regression (LR). After evaluating the proposed hybrid method via recall, precision, F1, and accuracy, our proposed method yielded satisfactory outcomes at 0.820, 0.805, 0.810, and 0.815, respectively. Furthermore, we also compared our hybrid method to a baseline utilizing the same training and test sets. The recall and precision scores of our proposed method were marginally higher than the baseline, with enhanced recall and precision scores at 4.76% and 4.88%, respectively.
<br>Citations: 1
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: Automatically Correcting Noisy Labels for Improving Quality of Training Set in Domain-specific Sentiment Classification
<br>Cover Date: 2023-01-01
<br>Cover Display Date: 2023
<br>DOI: 10.55003/cast.2022.02.23.006
<br>Description: Classification model performance can be degraded by label noise in the training set. The sentiment classification domain also struggles with this issue, whereby customer reviews can be mislabeled. Some customers give a rating score for a product or service that is inconsistent with the review content. If business owners are only interested in the overall rating picture that includes mislabeling, this can lead to erroneous business decisions. Therefore, this issue became the main challenge of this study. If we assume that customer reviews with noisy labels in the training data are validated and corrected before the learning process, then the training set can generate a predictive model that returns a better result for the sentiment analysis or classification process. Therefore, we proposed a mechanism, called polarity label analyzer, to improve the quality of a training set with noisy labels before the learning process. The proposed polarity label analyzer was used to assign the polarity class of each sentence in a customer review, and then polarity class of that customer review was concluded by voting. In our experiment, datasets were downloaded from TripAdvisor and two linguistic experts helped to assign the correct labels of customer reviews as the ground truth. Sentiment classifiers were developed using the k-NN, Logistic Regression, XGBoost, Linear SVM and CNN algorithms. After comparing the results of the sentiment classifiers without training set improvement and the results with training set improvement, our proposed method improved the average scores of F1 and accuracy by 20.59%.
<br>Citations: 2
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: Predicting the Close-price of Cryptocurrency Using the Kernel Regression Algorithm
<br>Cover Date: 2023-01-01
<br>Cover Display Date: 2023
<br>DOI: 10.1109/RI2C60382.2023.10356032
<br>Description: The aim of this work is to utilize the kernel regression (KR) approach to predict the closed-price for cryptocurrencies. This study makes use of three datasets: Bitcoin (BTC), Litecoin (LTC), and Ethereum (ETH). The min-max normalization method was used to scale feature values to a common range, often between 0 and 1. Furthermore, support vector regression (SVR) and long-short term memory (LSTM) were used to compare the prediction model-based on KR. The result of the KR models utilizing RMSE and MAPE demonstrated that the predictive model-based on KR gave more satisfying results.
<br>Citations: 1
<br>Aggregation Type: Conference Proceeding
<br>-------------------
<br><br><br>Title: ASPECT-BASED SENTIMENT CLASSIFICATION FOR CUSTOMER HOTEL REVIEWS
<br>Cover Date: 2022-12-01
<br>Cover Display Date: December 2022
<br>DOI: 10.24507/icicelb.13.12.1291
<br>Description: Using only ratings to gauge public opinion about products and services is in-sufficient to improve product quality or understand the reasons for consumer preferences. This problem was addressed by employing feature/aspect-based sentiment analysis to ex-amine the polarity of customer evaluations. An aspect-based sentiment analysis method was designed for hotel evaluations, taking account of staff attentiveness, room cleanliness, hotel facilities, value for money and location convenience. A collection of keywords for each hotel aspect was learned using Word2Vec as one of the three fundamental solution mechanics. This corpus was then utilized to select hotel features during developing an aspect-based multiclassification model to categorize sentences containing customer evaluations into their specific aspect classes. A binary-based sentiment classifier was also developed to assign the sentiment polarity of each sentence in each aspect class. Term frequency-inverse gravity moment (tf-igm) was employed as a term weighting scheme, while the SVM algorithm was used to construct text classification models. Our proposed method gave superior results to the baseline with improved average recall, precision, F1 and accuracy scores of 3.45%, 2.38%, 2.35% and 2.35%, respectively, compared to the baseline.
<br>Citations: 1
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: AUTOMATICALLY IDENTIFYING OF PLAGIARIZED SUBJECTIVE ANSWERS FOR THAI USING TEXT-BASED SIMILARITY ANALYSIS METHOD
<br>Cover Date: 2022-06-01
<br>Cover Display Date: June 2022
<br>DOI: 10.24507/icicel.16.06.639
<br>Description: In the context of education, many researchers design and develop methods or tools to identify plagiarism and maintain study quality. Text-based plagiarism often occurs in the academic domain, including online subjective examinations. Each one of the numerous proposed techniques has limitations in plagiarism detection. Here, a method is presented to identify plagiarized subjective answers in Thai when the subjective examination is performed online using natural language processing techniques (e.g., POS tagging) and cosine similarity analysis. The proposed method is called “similarity analysis of linguistic syntax and words used”. Results gave scores of true positive rate (TPR) as 0.81. Furthermore, the proposed method was compared with the baseline and when compared to the baseline, our proposed method improved the average TPR by 7.69%. This may demonstrate the success of our proposed method in identifying plagiarized subjective answers.
<br>Citations: 0
<br>Aggregation Type: Journal
<br>-------------------
<br><br><br>Title: Feature Extraction Efficient for Face Verification Based on Residual Network Architecture
<br>Cover Date: 2021-01-01
<br>Cover Display Date: 2021
<br>DOI: 10.1007/978-3-030-80253-0_7
<br>Description: Face verification systems have many challenges to address because human images are obtained in extensively variable conditions and in unconstrained environments. Problem occurs when capturing the human face in low light conditions, at low resolution, when occlusions are present, and even different orientations. This paper proposes a face verification system that combines the convolutional neural network and max-margin object detection called MMOD + CNN, for robust face detection and a residual network with 50 layers called ResNet-50 architecture to extract the deep feature from face images. First, we experimented with the face detection method on two face databases, LFW and BioID, to detect human faces from an unconstrained environment. We obtained face detection accuracy &gt; 99.5% on the LFW and BioID databases. For deep feature extraction, we used the ResNet-50 architecture to extract 2,048 deep features from the human face. Second, we compared the query face image with the face images from the database using the cosine similarity function. Only similarity values higher than 0.85 were considered. Finally, the top-1 accuracy was used to evaluate the face verification. We achieved an accuracy of 100% and 99.46% on IMM frontal face and IMM face databases, respectively.
<br>Citations: 2
<br>Aggregation Type: Book Series
<br>-------------------
<br><br><br>Title: Comparative Study between Texture Feature and Local Feature Descriptors for Silk Fabric Pattern Image Recognition
<br>Cover Date: 2020-03-19
<br>Cover Display Date: 19 March 2020
<br>DOI: 10.1145/3388176.3388201
<br>Description: Thai silk fabrics have unique patterns in different regions of Thailand. The designers may have been inspired and took ideas from the natural environment to create new silk patterns. Hence, many new silk patterns are modified from the original silk pattern. It is challenging for people to recognize a pattern without any prior knowledge and expertise. This paper aims to present a comparative study between texture feature and local feature descriptor for silk pattern image recognition. First, two feature extraction techniques: texture feature and local feature descriptors are proposed to create robustness features from sub-regions that are divided by the grid-based method. Second, the robust features are then classified using the well-known and effective classifier algorithms: K-nearest neighbor (KNN) and support vector machine (SVM) with the radial basis function. We experimented with silk pattern image recognition on two silk fabric pattern image datasets: the Silk-Pattern and Silk-Diff-Pattern. The evaluation results show that the texture feature called the local binary pattern (LBP) when combined with the KNN and SVM algorithms outperforms other feature extraction methods, even deep learning architectures.
<br>Citations: 0
<br>Aggregation Type: Conference Proceeding
<br>-------------------
<br><br><br>Title: Gender Recognition from Facial Images using Local Gradient Feature Descriptors
<br>Cover Date: 2019-10-01
<br>Cover Display Date: October 2019
<br>DOI: 10.1109/iSAI-NLP48611.2019.9045689
<br>Description: Local gradient feature descriptors have been proposed to calculate the invariant feature vector. These local gradient methods are very fast to compute the feature vector and achieved very high recognition accuracy when combined with the support vector machine (SVM) classifier. Hence, they have been proposed to solve many problems in image recognition, such as the human face, object, plant, and animal recognition. In this paper, we propose the use of the Haar-cascade classifier for the face detection and the local gradient feature descriptors combined with the SVM classifier to solve the gender recognition problem. We detected 4, 624 face images from the ColorFERET dataset. The face images data used in gender recognition included 2, 854 male and 1, 770 female images, respectively. We divided the dataset into train and test set using 2-fold and 10-fold cross-validation. First, we experimented on 2-fold cross-validation, the results showed that the histogram of oriented gradient (HOG) descriptor outperforms the scale-invariant feature transform (SIFT) descriptor when combined with the support vector machine (SVM) algorithm. The accuracy of the HOG+SVM and the SIFT+SVM were 96.50% and 95.98%. Second, we experimented on 10-fold cross-validation and the SIFT+SVM showed high performance with an accuracy of 99.20%. We discovered that the SIFT+SVM method needed more training data when creating the model. On the other hand, the HOG+SVM method provided better accuracy when the training data was insufficient.
<br>Citations: 8
<br>Aggregation Type: Conference Proceeding
<br>-------------------
<br><br><br>Title: Recognizing pornographic images using deep convolutional neural networks
<br>Cover Date: 2019-04-15
<br>Cover Display Date: 15 April 2019
<br>DOI: 10.1109/ECTI-NCON.2019.8692296
<br>Description: In this paper, we propose to use deep convolutional neural network (CNN) architectures, namely the deep residual networks (ResNet), the GoogLeNet, the AlexNet, and the AlexNet architectures, for pornographic image dataset. Also, the local descriptors, namely the local binary patterns (LBP), the histogram of oriented gradients, and the scale invariant feature transform (SIFT) combined with a support vector machine (SVM), a multilayer perceptron (MLP), or a K-nearest neighbor (KNN) techniques are proposed. Additionally, a bag of visual words (BOW) and the BOW using extracted HOG features (HOG-BOW) are compared. To classify the pornographic images, we compare the CNN architectures to well-known local descriptor techniques combined with the SVM, the MLP, and the MLP methods. Experimental results indicate that the ResNet architecture yields higher accuracies than all other approaches.
<br>Citations: 14
<br>Aggregation Type: Conference Proceeding
<br>-------------------
<br><br><br>Title: Fair payoff distribution in multiagent systems under Pareto optimality
<br>Cover Date: 2017-07-01
<br>Cover Display Date: 1 July 2017
<br>DOI: 10.1109/INCIT.2017.8257861
<br>Description: This research proposes a set of algorithms to compute fair payoff distribution among agents in service composition domain based on their contribution. In our system, intelligent agents, representing service providers, negotiate among themselves and form composite services to satisfy multiple-objective requirements. The quality of service for each objective is measured in term of degree of satisfaction. The overall quality of service is achieved by maximizing requesters satisfaction on all objectives according to Pareto optimality. We then deploy Shapley Value concept for fair payoff distribution among agents based on their contributions to the requesters optimal satisfaction. Since the computational complexity for Shapley Value is exponential, we are interested in investigating how well the algorithms for computing payoff perform. We found that on a typical computer, the algorithm can cope with around 20 agents with reasonable computational time.
<br>Citations: 0
<br>Aggregation Type: Conference Proceeding
<br>-------------------
<br><br><br>