Sensor Technology Box (Current)
We are looking into developing a sensor technology box for assisted living. This sensor technology box is not another smart home in a box product. Our primary concern is to identify and predict activities of daily living using ambient sensors. From research perspective, we are devising an algorithm for multi inhabitant environment and from technical aspect, our primary goal is to design the system in a plug and play manner like OSGI framework.
Active Learning Enabled Activity Recognition (Current)
Activity recognition in smart environment has been investigated rigorously in recent years. Researchers are enhancing the underlying activity discovery and recognition process by adding various dimensions and functionalities. But one significant barrier still persists which is collecting the ground truth information. Ground truth is very important to initialize a supervised learning of activities. Due to a large variety in number of Activities of Daily Living (ADLs), acknowledging them in a supervised way is a non-trivial research problem. Most of the previous researches have referenced a subset of ADLs and to initialize their model, they acquire a vast amount of informative labeled training data. On the other hand to collect ground truth and differentiate ADLs, human intervention is indispensable. As a result it takes an immense effort and raises privacy concerns to collect a reasonable amount of labeled data. We investigate and analyze different active learning strategies to scale activity recognition.
Following healthy lifestyle is a key for active living. Regular exercise, controlled diet and sound sleep play an invisible role on the wellbeing and independent living of the people. Sleep being the most durative activities of daily living (ADL) has a major synergistic influence on people’s mental, physical and cognitive health. Understanding the sleep behavior longitudinally and its underpinning clausal relationships with physiological signals and contexts (such as eye or body movement etc.) horizontally responsible for a sound or disruptive sleep pattern help provide meaningful information for promoting healthy lifestyle and designing appropriate intervention strategy. We propose to detect the microscopic states of the sleep which fundamentally constitute the components of a good or bad sleeping behavior and help shape the formative assessment of sleep quality. We initially investigate several classification techniques to identify and correlate the relationship of microscopic sleep states with the overall sleep behavior.
Monitoring road surface anomalies is important for both driver and road maintainers perspective to ensure a safer and comfortable transportation environment. Many of these road anomalies causes unpredictable occurrences for drivers and pedestrians. To maintain the road conditions, the authorities have to identify the anomalies first. For large scale deployment of road surface monitoring participatory sensing will prove to be vital and effective. We are devising a model based on mobile platform to assess the road surface monitoring especially the potholes. We use the accelerometer sensor of the smartphones which ensures both the cost effectiveness and multiple collaboration of the cars.
Energy Efficient Routing in Wireless Sensor Network for Multi-commodity Based Network
During my undergraduate thesis I did research on energy efficient routing protocols in sensor network and found those very challenging as there are so many constraints to consider. I thoroughly studied the important and prominent routing protocols and their limitations. The scheme I proposed involved data aggregation along the path. Data aggregation technique is invoked by building a merged tree from the distributed trees for overlapping paths along the network. Readings for the same target node are buffered along the path which ensured fewer collisions in the network. I also noted although data aggregation results in fewer transmissions, there is a tradeoff potentially greater delay in the case of some aggregation functions because data from nearer sources may have to be held back at an intermediate node in order to be aggregated with data coming from sources that are farther away. We simulated our algorithm with Network Simulator 2 (NS-2) with Mannasim framework. We published our work in “Workshop on Wireless Network and Communication (WNC)” BUET, 2013.
Artificial Agent for Better Banking
In most of the cases the banking analysts analyze the data based on their past experience and use some discrete tools which we found during our market research. We talked with some top banking professionals and shared our ideas and views. I with my other team mates developed a banking system and participated in a Software development competition (CITI Financial IT Case Competition 2011) when I was in my fourth year, “Artificial Agent for Better Banking” that utilizes data mining techniques for better analyzing banking data and plot graphs. During the development process we had insufficient amount of training data and they were low in quality, so it was difficult to apply clustering process. We used K-Means algorithm for clustering the data and then applied regression techniques to plot the graphs. We used RFM (Recency, Frequency and Monetary) score of the customers for marketing segmentation. Apart from winning the prize, it was a good practice of the incremental approach of development for me.