Research/Blog
CellStrat > Research/Blog > Artificial Intelligence > Computer Vision > CellStrat AI Lab – Minutes from 29th Jan 2019 BLR Meetup
CellStrat AI Lab – Minutes from 29th Jan 2019 BLR Meetup
- January 30, 2019
- Posted by: vsinghal
- Category: Computer Vision Deep Learning Natural Language Processing
No Comments
![CellStrat AI Lab Meetup - 29th Jan 2019, BLR](http://learning.cellstrat.com/wp-content/uploads/2019/01/Collage-2-840x430.png)
The AI Lab group continues to grow and more and more cutting-edge areas of Artificial Intelligence and Machine Learning are being researched and worked on.
In this meetup, the AI Lab teams discussed :-
- Semantic Image Segmentation with the U-Net architecture
- U-Nets are one kind of Encoder-Decoder Architecture to solve Semantic Image Segmentation. This architecture was proposed by Ronneberger et al (https://arxiv.org/pdf/1505.04597.pdf ) and it is particularly useful for biomedical image segmentation, which are generally hard to classify.
- U-Nets use Skip Connections from Encoder layers to Decoder layers – it that sense they are similar to Fully Convolutional Networks, or FCN, another type of Image Segmentation network when they are used with Skip connections.
- This network basically uses data augmentation to use the available data more efficiently. In that sense, it can be trained with relatively few images.
![](http://www.cellstrat.com/wp-content/uploads/2019/01/U-Net.png)
- Movie Recommender using Deep Conversational dataset
- ReDial is a dataset with 10000 conversations around the theme of movie recommendations. This was developed via use of Amazon Mechanical Turk (AMT) micro-task agent service. Here two agents are assigned to discuss movie recommendation with one party asking the other about movie recommendations.
- This uses a complex Encoder-Decoder GRU RNN scheme to recommend movies.
- Image Credit : http://papers.nips.cc/paper/8180-towards-deep-conversational-recommendations
![](http://www.cellstrat.com/wp-content/uploads/2019/01/Network.png)