Research/Blog
CellStrat > Research/Blog > Artificial Intelligence > CELLSTRAT AI LAB – MINUTES FROM 16TH. MAR 2019 BLR MEETUP
CELLSTRAT AI LAB – MINUTES FROM 16TH. MAR 2019 BLR MEETUP
- March 18, 2019
- Posted by: CellStrat Editor
- Category: Artificial Intelligence
No Comments
An intense AI Lab session took place on last Saturday in Bengaluru. Session saw participation of 32+ attendees of which 7-8 attended via Hangout from across th globe. Almost 10 folks were new and many have shown interest in attending future sessions.
- A marvellous presentation was made by Anupam Ranjan on
State of the Art pre-trained NLP model – Google’s BERT. He explained the concepts like transfer learning, Bidirectional Context, Transformer architecture. The session took a deep dive into core concepts of BERT – Masked Language Model(MLM), Next Sentence Prediction(NSP) and the BERT architecture. The session concluded with explanation of hyper parameter tuning of BERT and applications of BERT.
![](http://www.cellstrat.com/wp-content/uploads/2019/03/image.png)
- Anshumaan Dash explained about ANNs and CNNs in an intuitive manner.
- He gave a comprehensive analysis of CNN based architectures from AlexNet to DenseNet.
- He also gave a thorough explanation of residual blocks in ResNets, with focus on skip connections, (1×1) convolutions and change in learning objective.
![](http://www.cellstrat.com/wp-content/uploads/2019/03/160319_Residual-Learning_Anshumaan.jpeg)
- Vivek Singhal spoke about CellStrat’s Mission 10X project.
- The AI Lab group continues its learnings and research on high-tech topics. Attendees present already asked for deeper sessions by presenters on BERT, CNN etc. in next set of sessions.
![](http://www.cellstrat.com/wp-content/uploads/2019/03/CAILB_160319-Collage.png)