Registered Data

[CT121]

[00310] Human Activity Recognition from Inertial Motion Data

  • Session Date & Time : 2E (Aug.22, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : Human activity recognition (HAR) using inertial motion streaming has gained a lot of momentum in recent years. This has been driven by smart environments and the ubiquity of inertial-motion sensors in modern commodity devices. HAR applications span all aspects of human life such as healthcare, sports, manufacturing, etc. In this talk we give a brief description of the state-of-the-art work in HAR including action recognition, biometrics analysis (gender, age,..), sensor’s location determination, gait analysis, etc.
  • Classification : 68T01, 68T05, 92C47
  • Author(s) :
    • Walid Gomaa (Egypt Japan University of Science and Technology)

[00916] Deep Learning Approach Combined with Simulation As a Service to Generate Simulation from Sketched Image

  • Session Date & Time : 2E (Aug.22, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : A Simulation environment is an integrated software combining various interactive objects into one graphical user interface. This paper presents a new approach to generate simulation software from simulation sketched image. We use deep learning method based on (CNNs) and (RNN) with feature extractors MobileNet and ResNet for recognition combined with a Simulation as a Service (SaaS) to provide services for detected objects. We evaluated our approach on “Tu-berlin” dataset, 80% of accuracy was achieved.
  • Classification : 68T01, 68T45, 68T50, Deep Learning, Machine vision and scene understanding
  • Author(s) :
    • Mohamed Serrhini (University Mohamed Premier Oujda Morocco)

[01035] Reinforcement learning-based routing strategy in IoT applications using MDC

  • Session Date & Time : 2E (Aug.22, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : WSNs and IoT devices consume more power for data transmission. To reduce energy consumption, most of the traditional learning methodologies need enormous volumes of data and feature engineering, thus raising the learning complexity. A reliable reinforcement learning-based MDC model for effective routing is proposed to lower the learning complexity. Furthermore, the Q-Learning methodology is used to enhance learning along the shortest path. Combining these techniques can improve network stability while also enhancing routing performance significantly.
  • Classification : 68T01, 68T07, 68T35, Reinforcement learning, Machine Learning
  • Author(s) :
    • Muralitharan Krishnan (Sungkyunkwan University)
    • Yongdo Lim (Sungkyunkwan University)

[01078] DNN-based hybrid ensemble learning strategy for XSS detection and defense

  • Session Date & Time : 2E (Aug.22, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : Due to the high level of intelligence displayed by attackers, existing web-based security applications have failed. When attackers make changes to an organization's data, it is one of the most dangerous attacks (XSS). Combining ML and DL frameworks is proposed as a way to detect and defend against XSS assaults with high accuracy and efficiency. Using this representation, a new method is developed for integrating stacking ensembles into web-based software, which is called "hybrid stacking".
  • Classification : 68T01, 68T05, 68T07, Machine Learning, Deep Learning
  • Author(s) :
    • Seethalakshmi Perumal (MIT Campus, Anna University - Chennai)

[00315] Motion Assessment in Human Action Performance

  • Session Date & Time : 2E (Aug.22, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : Elderly people can be provided with safer independent living by the early detection of abnormalities in their motion actions performance. Low-cost depth sensing is one of the emerging technologies that can be used for unobtrusive and inexpensive motion abnormality detection and quality assessment. In this study, we developed and evaluated vision-based methods to detect and assess neuromusculoskeletal disorders manifested in common daily activities using three-dimensional skeletal data provided by the SDK of depth camera.
  • Classification : 68T05, 68T01, 92C47
  • Author(s) :
    • Walid Gomaa (Egypt Japan University of Science and Technology)