Techshows & Posters

Techshows

  1. Alt Inc.
    • Phần mềm Trí tuệ nhân tạo cá nhân al+
  2. AVA Vision
  3. Emotiv
    • EMOTIV Insight 5 Channel Mobile EEG
  4. Financial Deep Mind
  5. Five9
    • IBM Watson for Oncology
    • Word Embedding trong Tiếng Việt
  6. FPT Corporation
    • FPT Digital platform
  7. GraphicsMiner Lab
    • Force feedback interface – Haptics
  8. MICA
    • Home appliances control by hand gesture
    • 3D object detecting and fitting
    • Vietnamese medicinal plant retrieval
    • VIVA (VIetnamese Voice Assistant on smartphone)
  9. Ohmnilabs
    • Ohmni Robots
  10. SoICT
    • Hệ thống tổng hợp, phân tích thị trường bất động sản trực tuyến giasan.vn
    • Giải pháp chuyển đổi từ văn bản thành tiếng nói tiếng Việt tự nhiên vbee.vn
    • Hệ thống nhận dạng khuôn mặt BKFace
  11. TTV
  12. University of Engineering and Technology
    • DoIT: Hệ thống hỗ trợ nâng cao chất lượng tài liệu
    • Cờ toán Việt Nam
  13. VAIS
    • Hệ thống nhận dạng tiếng nói tiếng Việt
  14. VCCorporation
  15. Viettel Research & Development Institue
    • Thiết bị báo hiệu cứu bạn cá nhân VPLB
  16. VNG Corporation
    • AI Chatbot
    • Face Check-in
  17. VP9 Corporation

Posters

Tran Viet Trung
Harnessing the power of data for Vietnam real-estate market
Real estate is one of the most potential market in the world, no matter the economic condition. Our objective is to build an online data-driven real estate marketplace which are beneficial to both buyers and agents. We are harnessing the power of data to provide market trends, details ad-hoc analytics reports and client/potential agent matching. Technically, we are collecting huge amounts of data from public and trusted real estate listing sources. Then we put and restore order in complex datasets, cleaning, refining and structuring then in order to extract all relevant information. Finally, we provide various analytic reports in different views of real estate market in Vietnam.
Pham Duc Cuong, Fox Gregory, Marks Guy, Nguyen Anh Thu, Binh Nguyen, Hoa Nghiem, Loi Nguyen
Application of information communication technology in tuberculosis research – Opportunity for advancing tuberculosis program management
In 2016, the National Tuberculosis Program managed nearly 106 thousand patients with Tuberculosis (TB). Expanding capacity in the field of information and communication technology (ICT), particularly mobile technologies, brings new opportunities to improve TB management and control. However, the use of ICT in TB program is still limited. We review and report our experience in the application of mobile technologies to advance the management and implementation of TB research, with potential application to management of the TB program.
Le Dinh Thanh and Soichi Watanabe
Electric Field Estimations for Compliance Tests of Multiple-Antenna Transmitters in Wireless Communication Systems
Portable devices in the next generations of wireless communications will utilize multiple-antenna techniques, where the antennas work at a same frequency, for enhancing data rate. In compliance tests for safety usage of electromagnetics, there is an urgent demand of new methods to evaluate the power absorbed by a biological body when it is exposed to the electromagnetic emitted from such devices. In this study, we investigate in simple techniques to estimate the electric fields of multiple-antenna transmitters
Pham Van Toan, Nguyen Thanh Hau, Ta Minh Thanh
Deep learning ASR-based approach to non-native learner mispronunciation detection
Mispronunciation and accent detection based on the computer techniques are receiving increased attention nowadays. It is the most important part of second language acquisition. It can help non-native speakers to identify errors, to learn sounds with vocabularies, and to improve pronunciation performance. Along with the development of deep learning – DL, many applications have been proposed for solving such of issues. So many traditional problems such as computer vision, speech recognition, natural language processing can be resolved with higher accuracy when apply DL methods. In our works, we try some models of deep learning e.g CNN, RNN and those of combination for the phonetic classification in Japanese. Our work is applied in Chatty Pheasant – a mispronunciation detection mobile application of Framgia Incorporation.
Bui Thanh Hung
A Facial-expression Monitoring System to Recognize Students’ Engagement in the Classroom
An automatic facial expression recognition has been attracting more and more attention in the research area. It involves not only computer vision, machine learning but also behavioral sciences, and is used in many applications including but not limited to security, human-computer interaction, driver safety, healthcare and education. In this paper, we focus on the facial expression to recognize student’s engagement in classroom. we proposed how to recognize student’s facial expression in the classroom by using AAM and CNN model. We conducted experiment on CK+ database and evaluated by ROC, confusion matrix score and manual calculation of average emotions of the student in the class. We will develop this system to recognize faces and facial expressions of all students in the classroom and improve the system for more efficiency and better accuracy in both facial expression and emotion detection.
Tran Cao Truong, Bui Thu Lam
An Effective and Efficient Approach for Classification with Incomplete Data
Many real datasets suffer from the unavoidable issue of missing values. Classification with incomplete data has to be carefully handled because inadequate treatment of missing values will cause large classification errors. This paper proposes an effective and efficient approach to classification with incompleteness.
H Ruda Nie
Latent Semantic Analysis as a feature extraction technique for Text Categorization
Applying traditional machine learning classifiers to text categorization requires encoding text document into numerical vectors, which leads to two main problems in text categorization: 1. The huge dimensionality of features, and; 2. The sparse distribution in the data. Therefore, learning textual dataset by machine learning classifiers suffers the expensive computational cost, the high requirement of the computer memory size, and poor performance. To address these problems while classifying two datasets: Fake versus Real news (balanced dataset) and SMS spam collection (imbalanced dataset), this research applies Latent Semantic Analysis (LSA) based on Singular Value Decomposition (SVD) to reduce high dimension before utilizing Support Vector Machine to learn the data sets. The result shows that LSA using SVD effectively transforms high dimensional space into lower space without much loss of important information in the data. Thus, it significantly improves the performance of SVM on classifying Fake versus Real news, and SMS spam collection datasets.