A5070638524- Anomaly Detection Techniques and Applications
- Time Series Analysis and Forecasting
- Music and Audio Processing
- Complex Systems and Time Series Analysis
- Gear and Bearing Dynamics Analysis
- Water Quality Monitoring Technologies
- Machine Fault Diagnosis Techniques
- Advanced machining processes and optimization
- Vehicle License Plate Recognition
- Video Surveillance and Tracking Methods
- Handwritten Text Recognition Techniques
- Advanced Fiber Optic Sensors
- Traffic Prediction and Management Techniques
- Data Stream Mining Techniques
- Human Pose and Action Recognition
- Advanced Neural Network Applications
Tata Consultancy Services (India)
2021-2025
International Islamic University Chittagong
2017
Identification of user transport mode using mobile phone-based sensors is a key component Intelligent Transportation System. However, collecting labels/annotations while switching multiple modes into different journeys tedious. Also, type identification working across cities and countries prime need. This paper proposes method for generalizable journey detection without any annotations during training exploiting unsupervised representation learning. Our uses commonalities diversities various...
Traffic Sign Recognition (TSR) system is a component of Driving Assistance System (ADAS). The TSR assists the drivers in safe driving as road signs provide important information road. This research focuses to design and develop by using color cues Convolution Neural Network (CNN) both features extractor classifier for Bangladeshi traffic signs. In first step, after image acquisition, some pre-processing task performed. Then segmented HSV model. After that, morphological closing executed fine...
Getting a robust time-series clustering with best choice of distance measure and appropriate representation is always challenge. We propose novel mechanism to identify the clusters combining learned compact time-series, Auto Encoded Compact Sequence (AECS) hierarchical approach. Proposed algorithm aims address large computing time issue as latent AECS has length much less than original at same want enhance its performance.Our exploits Recurrent Neural Network (RNN) based under complete...
Driving behavior monitoring plays a crucial role in managing road safety and decreasing the risk of traffic accidents. is affected by multiple factors like vehicle characteristics, types roads, traffic, but, most importantly, pattern driving individuals. Current work performs robust analysis capturing variations patterns. It forms consistent groups learning compressed representation time series (Auto Encoded Compact Sequence) using multi-layer seq-2-seq autoencoder exploiting hierarchical...
Time-series generated by end-users, edge devices, and different wearables are mostly unlabelled. We propose a method to auto-generate labels of un-labelled time-series, exploiting very few representative labelled time-series. Our is based on representation learning using Auto Encoded Compact Sequence (AECS) with choice best distance measure. It performs self-correction in iterations, latent structure, as well synthetically boosting time-series Variational-Auto-Encoder (VAE) improve the...
Intelligent continuous monitoring of an IoT system to identify the operational changes, encompassing both normal and abnormal scenarios, with drift in sensing device is a challenging problem. It demands capability learning continuously multiple interventions or shifts, without forgetting past events information. However, learned knowledge, known as catastrophic forgetting, impacts significantly on performance monitoring. In this work, we propose generative neural network based model handle...
In this work, we perform Physics guided data synthesis. Proposed method uses seed from the observed source state / domain for generation of unobserved target domain. Our adapts variation features with respect to across states. This approach knowledge and its associated impacts on statistical signal properties data. We use generative learning comprising a variational auto-encoder (VAE) based neural network model. model has influenced optimization function achieve adaptation domains. aims...
Today's world extensively depends on analytics of high dimensional sensor time-series, and, extracting informative representation. Sensor time-series across various applications such as healthcare and human wellness, machine maintenance etc., are generally unlabelled, getting the annotations is costly time-consuming. Here, we propose an unsupervised feature selection method exploiting representation learning with a choice best clustering recommended distance measure. Proposed reduces space,...