A5011986638- Groundwater and Watershed Analysis
- Anomaly Detection Techniques and Applications
- Remote Sensing and Land Use
- Face and Expression Recognition
- Gaze Tracking and Assistive Technology
- Reliability and Maintenance Optimization
- Soil and Land Suitability Analysis
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Hydrology and Watershed Management Studies
- Neural dynamics and brain function
- EEG and Brain-Computer Interfaces
- Material Properties and Processing
- Animal Nutrition and Physiology
- Geological Modeling and Analysis
- Probabilistic and Robust Engineering Design
- Urban Heat Island Mitigation
- Geophysics and Gravity Measurements
- Smart Grid Energy Management
- Bird parasitology and diseases
- Fluid Dynamics and Heat Transfer
- Simulation Techniques and Applications
- Materials Engineering and Processing
- Metaheuristic Optimization Algorithms Research
- Blind Source Separation Techniques
- Hydrology and Drought Analysis
Ho Chi Minh City University of Technology
2009-2023
Vietnam National University Ho Chi Minh City
2014-2023
University of South Florida
2022
North Dakota State University
2022
University of North Dakota
2022
An Giang University
2021
University of Ottawa
2014
Wilfrid Laurier University
2014
University of Canberra
2005-2012
National Institute of Occupational Health
2002
Low-rank adaptation (LoRA) has emerged as a powerful method for fine-tuning large-scale foundation models. Despite its popularity, the theoretical understanding of LoRA remained limited. This paper presents analysis by examining connection to Mixture Experts Under this framework, we show that simple reparameterizations matrices can notably accelerate low-rank matrix estimation process. In particular, prove reparameterization reduce data needed achieve desired error from an exponential...
Brain computer interface (BCI) is about the communication channel between brain of a human subject and computerized device. Electroencephalography (EEG) signals are primary choice as sources interpreting intention subject. EEG have long history being used in health for purposes studying activities medical diagnosis. very weak to contamination from many artifact signals. For applications health, true signals, without contamination, highly desirable. However, BCI, where stable patterns source...
In recent years, lithium-ion batteries (LIB) have been used widely in portable electronic devices because of their advantages durability, stability, high-capacity, low-cost, light-weight and smallscale. It makes LIB also deployed various complex systems, which efficient prediction battery data, especially state-of-health (SoH), becomes crucial to ensure that the systems work stably without risks power interruptions. With improvement Artificial Intelligence (AI) technologies, many works...
Support Vector Machine (SVM) is a well-known kernel-based method for binary classification problem. SVM aims at constructing the optimal middle hyperplane which induces largest margin. It proven that in linearly separable case, this offers high accuracy on universal datasets. However, real world datasets often contain overlapping regions and therefore, decision should be adjusted according to profiles of In paper, we propose Robust (RSVM), where hyperplanes can properly accommodate By...
One-class Support Vector Machine (OCSVM) is a well-known method for novelty detection. However, OCSVM regards all negative data samples as common symbol and thereby not being able to utilize the information carried by them. Furthermore, requires fully labeled set cannot work efficiently with both unlabeled which very popular nowadays. In this paper, we first extend model of enable using samples. We then propose two methods integrate semi-supervised learning paradigm extended detection purpose.
Support Vector Data Description (SVDD) is known as one of the best kernel-based methods for one-class classification problems. SVDD requires fully labelled data sets. However, in reality, an abundant amount can be easily collected, while labelling process often expensive, time-consuming, and error-prone. Therefore, partially sets are popular easy to obtain. In this paper, we propose a semi-supervised learning method, Fuzzy Entropy Semi-supervised (FS3VDD), extend cope with The model employs...
One of the important problems in medical imaging is two-class classification, for example determination benign from malignant cases breast cancer treatment. In this paper we present a new support vector machine method image classification. The key idea to construct an optimal hypersphere such that both interior margin between surface sphere and normal data, exterior abnormal data are as large possible. proposed easily implemented can reduce false positive negative error rates obtain very...
Water resources vulnerability assessment provides the basis for policymakers and stakeholders regarding state of river basin to be focused its better management. This research aims assess index (VI) upper part Dong Nai (UPDN) basin, which plays a crucial role socio-economic development Ho Chi Minh City 10 other provinces in Southeast Vietnam, as well sub-basins. study accounts from year 1994 2030 (36 years) by using an integrated solution between GIS, remote sensing (RS) SWAT model...
Support vector data description (SVDD) has been widely used in pattern classification, however it does not provide high performance brain-computer interface (BCI) classification problems since brain signals are noisy and chaotic. Brain have distinct distributions hence a hyper-sphere SVDD could well describe the data. We propose this paper multi-sphere approach to better for also fuzzy clustering optimize parameters. Experiments on set III motor imagery problem BCI Competition II were...
Modeling neural population dynamics underlying noisy single-trial spiking activities is essential for relating observation and behavior. A recent non-recurrent method - Neural Data Transformers (NDT) has shown great success in capturing with low inference latency without an explicit dynamical model. However, NDT focuses on modeling the temporal evolution of activity while neglecting rich covariation between individual neurons. In this paper we introduce SpatioTemporal Transformer (STNDT),...
Abstract Sustainable management of the river basin is a profound challenge for environmental in context climate change. Drought situations occur relation to meteorological, hydrological, agricultural factors and change as well. In this study, remote sensing technology was applied assess impacts on drought Ba River basin, Central Vietnam. has been created by land use/land cover changes recent years, which resulted sharp decrease forest area period 1989 2019 (-41.5%) significant increase with...
Decoding attempted speech from neural activity offers a promising avenue for restoring communication abilities in individuals with impairments. Previous studies have focused on mapping to text using phonemes as the intermediate target. While successful, decoding directly ignores context dependent nature of activity-to-phoneme brain, leading suboptimal performance. In this work, we propose use diphone - an acoustic representation that captures transitions between two context-aware modeling We...
Speech brain-computer interfaces aim to decipher what a person is trying say from neural activity alone, restoring communication people with paralysis who have lost the ability speak intelligibly. The Brain-to-Text Benchmark '24 and associated competition was created foster advancement of decoding algorithms that convert text. Here, we summarize lessons learned ending on June 1, 2024 (the top 4 entrants also presented their experiences in recorded webinar). largest improvements accuracy were...
Permanent Scatterer InSAR (PSInSAR) technique is used commonly to evaluate land subsidence for some cities in the world and pilot project Ho Chi Minh City has shown capability of this mapping period from 2006 2010. The result shows that PSInSAR can be applied study phenomenon more efficiently than any traditional methods. However, SAR data often are commercial images such as ERS, ALOS, COSMO SkyMed… a large number scenes must collected ensure best results. This paper introduces results...
Eimeria species are causative pathogens considered as a tremendous threat to the poultry industry in Vietnam.Sufficient assessment of coccidiosis prevalence is critical for improving prophylaxis strategies control disease.The objective current study was investigate local chickens (Noi breed) Mekong Delta, Vietnam.A total 1200 fecal samples were collected from 20 various farms Ben Tre and Hau Giang provinces.Using microscopic examination, rates provinces 65.83% 68.50%, respectively.Eimeria...
This paper introduces the solution for Dalat city’s soil erosion mapping using integration of GIS, Remote Sensing and Universal Soil Loss Equation (USLE). Each USLE factors with associated attribute data are dicussed parameters were selected encoded in a GIS database to produce thematic layers. The result demonstrates map that indicates potential annual loss located each area land. is used confirm severe level risk need immediate attention from conservation point view.
This paper describes a new approach for monitoring the construction progress of Urban Railway Construction Project “Metro line1 - Ben Thanh Suoi Tien” by using Unmanned Aerial Vehicles (UAV) to capture high resolution imagery at different stages project. The advantage AscTec Falcon 8 systems lies in their flexibility and efficiency capturing surface an area from low flight altitude. In addition, further information such as orthoimages, elevation models 3D objects can easily be processed...
According the national climate change scenarios, sea level rise causes not only flooding but also salt water intrusion and impacts severely on rice production. This paper demonstrates use integration solution of GIS for managing data, assessing due to forecasting salinity information at local scale Tra Vinh province. Toolkit set is established including 04 major functions: System administrator, Modeling, To create automatic reports Utilities updating. Results this study showed that proposed...
A high resolution remote imaging system using a vertical take-off and landing helicopter- type unmanned air vehicle has been built successfully tested. The aircraft is navigated differential GPS autonomously tracked set of waypoints defined on Google map. landing-type chosen due to its maneuverability capability flight in condensed area with physical obstacles. was planned taken pictures at low altitude changing from 10m 300m. experimental results shown that the image horizontal plan lower...