Research/Blog
CellStrat AI Lab – Minutes From 6th. Apr, 2019 Bangalore & Gurgaon Meetups
- April 8, 2019
- Posted by: CellStrat Editor
- Category: Artificial Intelligence Computer Vision Deep Learning Machine Learning
The intense AI and ML projects and discussions continue week after week. We had almost 25+ folks attend the lab in Bangalore and 10+ in Gurgaon including two visitors from New York and Melbourne with many more on the hangout.
Bangalore:
Vivek started with a discussion of Principal Component Analysis. PCA is a dimensionality reduction technique to reduce data dimensions such as 3D to 2D conversion. This helps reduce data complexity while maintaining most of data intelligence.
Anupam presented a superb session on Auto Text Summarization. He explained the difference between Extractive and Abstractive Summarization. Then he presented algorithmic details of Extractive Summarization, which involves discovering similarity patterns (using cosine similarity match) across multiple text lines and selecting sentences which display similar intent for summarization purposes. Anupam also presented an intuitive code demo for this solution.
![](http://www.cellstrat.com/wp-content/uploads/2019/04/image-3.png)
Finally Shreyas had an interesting session on DensePose algorithm (https://arxiv.org/abs/1802.00434), the cutting-edge Pose Estimation algo published by Facebook last year. Dense pose is used to predict the dense pose coordinates of a person. It uses a handful of algorithms such as maskrcnn, fpn, roialign, student teacher network, regression, cascading networks etc. DensePose involves a series of advanced computer vision concepts such as human annotations to UV map (a 2D mapping of 3D object such as a human body), Masked RCNN for image segmentation, ROI Align to find region of interests, regression to regress exact location of point on the UV map and also GANs to super-impose new persons on estimated pose model.
![](http://www.cellstrat.com/wp-content/uploads/2019/04/DensePose-Concpt_Shreyas_060419-1024x512.jpeg)
Team is also putting together a set of healthcare use cases in AI for upcoming partner sessions with NASSCOM. Status was reviewed on these use cases.
Industry portfolios are being worked upon for various sectors for corporate showcase. Currently teams are working on healthcare and driverless car use cases.
![](http://www.cellstrat.com/wp-content/uploads/2019/04/AILabB_060419_Collage.png)
Gurgaon:
Vinay presented how Classification of data can be done using Decision Trees. Various easy to understand examples were given and audience rated the explaination highly.
Pratima a good in-depth session on Python in Data Science and explained right from scratch for newbies in this line. It wa well understood and related by both learned and naiive in the audience.
![](http://www.cellstrat.com/wp-content/uploads/2019/04/image-2.png)
Dharmendra put forward a problem statement from industrial marketing perspective as he heads marketing for a large industrial equipments company. Audience showed good interest in this brainstorming session on how AI/ML can be used to solve the problem…everybody suggested that deeper understanding and discussion would be required for this and hence more sessions will follow in the future on this and future problems…
![](http://www.cellstrat.com/wp-content/uploads/2019/04/AILabG_060419-Collage.png)