KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 9 of 9

Some of the metrics are blocked by your
consent settings
Application of Natural Neighbor-based Algorithm on Oversampling SMOTE Algorithms
(2021-04-01)
Srinilta, Chutimet
;
Kanharattanachai, Sivakorn
Classification performance depends highly on data distribution. In real life, data often come imbalanced where one class is found more often than others. SMOTE-based algorithms are usually used to handle the class imbalance problem. One key parameter that algorithms in SMOTE family require is k-the number of nearest neighbors with respect to a certain data point. K that fits the dataset the most gives the optimum performance. This paper proposes an approach to suggest a value of the parameter k using Natural Neighbor algorithm. Datasets are made balanced by four SMOTE-based algorithms-standard SMOTE, Safe-Level-SMOTE, ModifiedSMOTE and Weighted-SMOTE. The F-measure and Recall matrices are used to evaluate classification performance of a Support Vector Machine classifier running against six datasets with different imbalance ratios. The results show that, the average classification performance achieved by the proposed k's is closer to the optimum when compared with the performance given by the default value of k.
Some of the metrics are blocked by your
consent settings
Multi-task learning and thai handwritten text recognition
(2020-07-01)
Srinilta, Chutimet
;
Chatpoch, Suchakree
Written languages have some commonalities and differences. Knowing one language can help learning other languages better. While labelled data is limited in offline Thai handwritten text recognition problem domain, multi-task learning is chosen to address the problem in our study. The multi-task model is trained from Thai, Latin and Devanagari writing scripts. The model consists of three layers employing Convolutional Neural Network, Recurrent Neural Network and Connectionist Temporal Classification, respectively. Recognition accuracies are compared against three corresponding single-task models. Thai handwritten text recognition performance is considerably improved by multi-task learning. Learning multiple languages helps the model generalize better when trained by large enough datasets.
Some of the metrics are blocked by your
consent settings
Preface
(2020-01-01)
Sitek, Paweł
;
Pietranik, Marcin
;
Krótkiewicz, Marek
;
Srinilta, Chutimet
Some of the metrics are blocked by your
consent settings
Municipal solid waste segregation with CNN
(2019-07-01)
Srinilta, Chutimet
;
Kanharattanachai, Sivakorn
Pollution from municipal solid waste has been a problem in Thailand for a long time. People generate waste in every minute. Ineffective waste segregation does increase difficulties in solid waste management. The Pollution Control Department of Thailand provides segregation guideline for municipal solid waste. Household wastes should be separated into four types-general waste, compostable waste, recyclable waste and hazardous waste. This paper explored performance of CNN-based waste-type classifiers (VGG-16, ResNet-50, MobileNet V2 and DenseNet-121) in classifying waste types of 9,200 municipal solid waste images. Waste type can be identified directly from waste-type classifier or derived from waste-item class. Derived classifiers outperformed their corresponding direct classifiers in the experiment. The highest waste-type classification accuracy was 94.86% from the derived ResNet-50 classifier.
Some of the metrics are blocked by your
consent settings
Machine learning approach in predicting post-transfusion packed cell volume in anemic dogs
(2018-08-13)
Srinilta, Chutimet
;
Sunhem, Wisuwat
;
Sangunwong, Pongsak
;
Chanchartree, Satthathan
Blood transfusion is commonly used to treat anemia. Blood transfusion is vital to life in many cases. Blood donation is a voluntary activity. In Thailand, blood supply for small animals are very limited. Therefore, blood must be used with extra care to save as many lives as possible. Success of whole blood transfusion where all blood components are transfused is determined by the rise of Packed Cell Volume (PCV) after transfusion. Veterinarians rely on formula to estimate the transfusion volume that can raise patient's PCV to the target. This paper attempted to use machine learning models to predict post-transfusion PCV in anemic dogs. Linear regression, XGBoost and Support Vector Regression algorithms were used in machine learning prediction models. Transfusion records from Kasetsart University Veterinary Teaching Hospital at Hua Hin were employed to assess model performance. The formula commonly used by veterinarians was performance comparison baseline. Wilcoxon signed-rank test was used to assess significant differences of the result. It was statistically confirmed with confidence interval of 90% that Support Vector Regression performed better than the baseline method on conventional input set alone and when certain red blood cell attributes were added to the conventional input set.
Some of the metrics are blocked by your
consent settings
Lyric-based Sentiment Polarity Classification of Thai Songs
(2017-01-01)
Srinilta, Chutimet
;
Sunhem, Wisuwat
;
Tungjitnob, Suchat
;
Thasanthiah, Saruta
;
Vatathanavaro, Supawit
Song sentiment polarity provides outlook of a song. It can be used in automatic music recommendation system. Sentiment polarity classification based solely on lyrics is challenging. It involves understanding linguistic knowledge, song characteristics and emotional interpretation of words. Since lyric is in a form of text. Techniques used in text mining, text sentiment analysis and music mood classification are studied and used together in our proposed model. Two types of classifier are proposed—lexicon-based classifier and machine learning-based classifier. N-gram model is used in feature set generation. Features are filtered by Information Gain. Feature weighting scheme is employed. We create a sentiment lexicon from Thai song corpus. Full lyric and certain parts of lyric are chosen for datasets. We evaluate our models under various environments. The best average accuracy achieved is 68%.
Some of the metrics are blocked by your
consent settings
Statistical downscaling for rainfall and temperature prediction in Thailand
(2011-07-26)
Aksornsingchai, Pawanrat
;
Srinilta, Chutimet
This paper studies three statistical downscaling methods to predict temperature and rainfall at 45 weather stations in Thailand. Methods under consideration are multiple linear regressions (MLR), support vector machine with polynomial kernel (SVM-POL), and support vector machine with Radial Basis Function kernel (SVM-RBF). Large-scale data are from Geophysical Fluid Dynamics Laboratory (GFDL). Five predictor variables are chosen: (1) temperature, (2) pressure, (3) precipitation, (4) evaporator, and (5) net short wave. Accuracy is assessed by 10-fold cross-validation in terms of root-mean-squared error (RMSE) and correlation coefficient (R). SVM-RBF is the most accurate model. Prediction accuracy of monthly average rainfall and temperature is satisfying in most part of the country. Lastly, downscaling models can project long term trends of monthly average rainfall and temperature.
Some of the metrics are blocked by your
consent settings
PP caching: Proxy caching mechanism for YouTube videos in campus network
(2011-07-26)
Suranuntakul, Varangpa
;
Srinilta, Chutimet
Nowadays, YouTube has become a very popular website offering video sharing service. YouTube accounts for a significant portion of Internet traffic in global and local networks. Many techniques have been implemented to deliver videos smoothly to every user. Streaming proxy servers are put into action to reduce user-perceived latency as well as network resource utilization. "What is viewed next" can be guessed from relationships of YouTube videos. This paper proposes PP caching mechanism that benefits from prefix caching and prefetching technique. Certain parts of selected videos are prefetched and kept at the proxy cache. The experimental results show that startup delay decreases and cache hit rate increases when the proposed mechanism is tested against actual YouTube traces from two campuses.
Some of the metrics are blocked by your
consent settings
Load balancing algorithms for internet video and audio server
(2001-01-01)
Niyato, Dusit
;
Srinilta, Chutimet
This paper reports the application of load balancing algorithms on Internet video and audio media server. We took into account two server architectures: centralized and distributed load balancing; and five load balancing algorithms: random, round-robin, adaptive bidding, diffusion, and state change broadcast. The startup time and system distance were measured in the experiments. The results show that the response time of distributed load balancing is less than the centralized load balancing architecture. And algorithms that use more load information can balance the load on video and audio media server better, but produce more response time. © 2001 IEEE.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results