#LTM - TensorFlow: CAD and Machine Learning

Agenda:

6:00 - Register and grab a drink (going through security may take longer than usual - please allow extra time for arriving)
6:30- Large-scale mammography CAD with deformable convolutional nets (Stephen Morrell)
7:00- Break time - networking, food & drinks!
7:30- Content Science: Machine Learning in the Newsroom (Dan Gilbert, News UK)
8:20- Wrap up & Networking

1. Stephen Morrell - Mammography and Transcription Specialist
   www.linkedin.com/in/stephenmorrell

Large-scale mammography CAD with deformable convolutional nets

State of the art deep learning methods for image processing are evolving into increasingly complex meta-architectures that incorporate a growing number of modules. Among them, deformable convolutional nets (DCN) and region-based fully convolutional networks (R-FCN) have the potential to revolutiose mammography CAD: R-FCN excels at object detection, while deformable convolution and pooling can model a wide range of mammographic findings of different shapes and scales, thanks to their versatility. In this study, we present a neural net architecture that adapts R-FCN / DCN to the mammographic image analysis problem. We motivate changes in architecture and parameters in order to bridge the gap between natural images and mammograms. We trained the network on a large, recently released dataset of mammograms (Optimam) including 6,500 cancerous images. By combining our modern architecture with such a rich dataset, we achieved an area under the curve of 87.9% for breast-wise detection in the DREAMS challenge (130,000 withheld images), which is higher than any other submission in the competitive phase of the challenge.

2. Dan Gilbert (@dangilbot) - Director of Data, News UK
   https://twitter.com/dangilbot

Content Science: Machine Learning in the Newsroom

The Data Technology team in News UK develops data platforms, machine learning systems and products. We work within our newsrooms to empower journalists and editors with access to data products that are carefully tuned to their needs. Dan will be discussing where we are on the journey, from the algorithms being developed to help aide decision making through the lifecycle of an article, to the modern APIs and front end frameworks being developed to provide access to the results of these algorithms.

-------

Thank you to our sponsors:

YLD
https://twitter.com/YLDio

YLD is one of London’s fastest growing software engineering consultancies building exciting products and platforms in React / Node for Trainline, Vogue, The Economists and many more. The team is committed the open source community aiming to create a long standing engineering culture and delivery capability in each piece of work.

YLD is looking for React / Node / DevOps engineers to join the team: https://jobs.lever.co/yld

Watch Tech Talks YLD youtube channel for more talks on React and Node:
www.youtube.com/TechTalksYLD

Join the conversation:

• Twitter @YLDio (https://twitter.com/YLDio)
• LinkedIn (www.linkedin.com/company/yld)

------

News UK (https://twitter.com/newsuktech)

News UK is a great company full of talented, dedicated and creative people. We are an organisation that holds journalism at its very heart. Our newspapers and digital products include some of the most powerful media brands in the English speaking world: the Times, The Sunday Times and The Sun - reaching 30 million people each week. Despite differences in audience and content our brands are united by a commitment to independent journalism that connects our customers.

------------------------------

***Please provide their full names when RSVP have a photo ID when entering the building. ***

Looking forward to seeing you soon,

#LTM Team