Hi everyone, Join us for the Change seminar tomorrow at 12:30 pm. In case you cannot attend in person, we will also live-stream the talk using the link https://youtu.be/4WVRKpvlFgs
Best, Samia Ibtasam <http://samiaibtasam.com/> PhD Student Paul G. Allen School of Computer Science & Engineering University of Washington On Mon, Mar 2, 2020 at 10:03 AM Samia Ibtasam <[email protected]> wrote: > Hi everyone, > > Please note that this week's special Change seminar will be on Thursday, > March 5th, 2020 starting at 12:30 pm. Please note the change in day and > time only for this week. There won't be a seminar on Tuesday this week. > > > *When: *Thursday 3/5, 12:30 pm-1:30 pm > > *Where:* CSE1 305 (Allen Building) > > *Who:* Dr. Mohsen Ali (Intelligent Machines Lab, Information Technology > University, Pakistan) > > *Title:* Deep Learning for Intelligent Mapping of Pakistan > > > *Abstract * > > The past decade saw the graduation of machine learning from the academic > setting to incorporation into the practical world. From self-driving cars > to designing antibiotics, from personal life to businesses we are relying > on the data-driven algorithms to make decisions for us. Dependence on the > data means that members of societies, which are not similar to the > technologically advanced ones, and which do not produce a regular and > uniform stream of data, are left behind. With a data democratization a > target, we are trying to fill the gap by designing deep learning-based > neural networks for satellite imagery to do urban analysis and combine the > information extracted from it with the existing geospatial data sources to > create rich informative maps of Pakistan in the public domain. > > > > In this talk, I will start by looking at an attention-based deep neural > network we designed to compute the fine-grain building foot-print map of > the province Punjab in Pakistan. Moving from there, I will talk about our > work on counting buildings from satellite imagery (think about boxes inside > boxes, next to the other boxes) and detecting areas devastated by war or > natural disaster. All this will lead to the Slum Detection project and how > we boot-strapped our algorithm to detect slums from satellite imagery in > three cities of Pakistan. And our current effort to use satellite imagery > and other GIS information to approximate multidimensional poverty. Finally, > I will introduce two on-going projects Tag-Pakistan and “Yadain: An Oral > History”; [Yadain means memories in Urdu] where we are using crowdsourced > images and augmented reality to create highly informative maps. > > > > > > *Bio:* > > Mohsen Ali is an Assistant Professor > <http://itu.edu.pk/faculty-itu/mohsen-ali/> at Information Technology > University and co-founder of the Intelligent Machines Lab > <http://im.itu.edu.pk/researches/>. IML has been established with the > objective to provide a platform for researchers and engineers working in > machine learning, computer vision, and robotics to collaborate and solve > real-world problems. Mohsen’s current interest lies in theoretical and > practical problems entailing Computer Vision and Machine Learning > (including Deep learning), and their application in the field of video and > image analysis, remote sensing and affective computing > <http://im.itu.edu.pk/research/affective-computing/>. His work on > decreasing the Affective Gap, between the multimedia content and modeling > emotional response, and designing Emotion Filters > <http://im.itu.edu.pk/affective-image-transfer/> to transform an image in > order to garner desired emotional impact, has been accepted in respected > computer vision conferences. > > > Currently, he is working on satellite imagery analysis using deep learning > to understand urbanization and economic condition. In collaboration with > UNDP and Sustainable Development Goal Tech Lab at ITU, he is working to > replace intensive surveying exercises to detect and estimate the economic > condition of slums, with satellite imagery and geospatial data analysis > techniques. His work on counting buildings > <http://im.itu.edu.pk/deepcount/> and detecting destroyed areas from > satellite imagery has been accepted in a top-ranking journal. Intending to > democratize the data, he is interested in using machine learning and > computer vision to collect large datasets capturing bio-diversity, culture, > and history to create an informative, self-growing and in-depth map of > Pakistan. His group is exploring techniques like active learning, domain > adaptation, unsupervised learning for the task. > > > Mohsen Ali completed his doctoral studies from the University of Florida > in the area of Computer Vision and is the recipient of the Fulbright > Scholarship award. His work has been accepted in venues like ICCV, CVPR, > ISPRS P&RS. > > Our Research Page: Computer Vision and Machine Learning Research Group > <http://im.itu.edu.pk/computer-vision-and-machine-learning-research-group/> > > Selected Publications: http://im.itu.edu.pk/publications/ > > Google Scholar Page: https://scholar.google.com.pk/ > <https://scholar.google.com.pk/citations?user=59ISSCEAAAAJ&hl=en> > > Best, > Samia Ibtasam <http://samiaibtasam.com/> > Ph.D. Student > Paul G. Allen School of Computer Science & Engineering > University of Washington > >
_______________________________________________ change mailing list [email protected] https://changemm.cs.washington.edu/mailman/listinfo/change
