Interpretable Machine Learning, Tweet Classifier, H2O World Highlights and More
Details
Dear All,
Notes:
On arrival, please sign in with your ID at the "East Desk" of One Canada Square and then proceed to 14th floor for the event.
We will have a professional videographer on site to record the talks. Videos will be uploaded to H2O's YouTube channels a few days after the meetup.
If you want to tweet about the event, please mention H2O's twitter handle @h2oai, Moody's handle @MoodysAnalytics and our meetup hashtag #LondonAI.
We are thrilled to see the growth of this meetup group (4500+ members right now). Let's end the year with one more exciting meetup. This time we will have speakers from Aviva, Theodo and Entrepreneur First.
Many thanks to our friends from Moody's Analytics, we have a super cool venue right in the heart of Canary Wharf. They also very kindly provide food and drinks for the event.
Agenda:
-
1800 to 1840 Doors open. Food/Drinks + Networking
-
1840 to 1900 Introduction by Joe (H2O.ai) and Tom (Moody's)
-
1900 to 1925 Kasia's Talk + Q&A
-
1925 to 1930 Short Break
-
1930 to 1955 Ben's Talk + Q & A
-
1955 to 2000 Short Break
-
2000 to 2025 Jack's Talk + Q & A
----------------------------------------------------------------
Kasia's Talk:
Kasia will discuss complexities of interpreting black-box algorithms and how these may affect some industries. She will present the most popular methods of interpreting Machine Learning classifiers, for example, feature importance or partial dependence plots and Bayesian networks. Finally, she will introduce Local Interpretable Model-Agnostic Explanations (LIME) framework for explaining predictions of black-box learners – including text- and image-based models - using breast cancer data as a specific case scenario.
About Kasia
Kasia Kulma is a Data Scientist at Aviva with a soft spot for R. She obtained a PhD (Uppsala University, Sweden) in evolutionary biology in 2013 and has been working on all things data ever since. For example, she has built recommender systems, customer segmentations, predictive models and now she is leading an NLP project at the UK’s leading insurer. In spare time she tries to relax by hiking & camping, but if that doesn’t work ;) she co-organizes R-Ladies meetups and writes a data science blog R-tastic (https://kkulma.github.io/).
LinkedIn: https://www.linkedin.com/in/kasia-kulma-phd-7695b923/
----------------------------------------------------------------
Ben's Talk:
Building a Trump/Obama Tweet Classifier with 98% accuracy in 1 hour!
When it comes to Machine Learning it can be hard to find small scale real world applications that deliver good results. To find such an application I challenged myself to build a Naive Bayesian tweet classifier. Classifying the author of any given tweet as President Obama or President Trump respectively, resulting in 98% accuracy!
This talk will go through scraping the data, building the classifier (without 'recoding the wheel') and showing how the model was verified.
Along with some live coding of the whole thing in JavaScript!
About Ben
I'm a developer, working with startups to launch MVP's and large corporates to deliver in startup speed. I work for Theodo, a London-Paris based startup, as part of a tech team for hire using cutting edge open source technology and a mature methodology to launch new products for clients.
I'm tech at heart, loving to code, participating in hackathons, guest lecturing as part of the University of Southampton CompSci course and acting as a tech advisor to multiple startups.
LinkedIn: https://www.linkedin.com/in/benjaminellerby/
----------------------------------------------------------------
Jack's Talk:
What people get wrong when building AI ventures
Entrepreneur First has more Machine Learning investments than anyone else in Europe. And they made those investments at incredibly early stage, backing people before they had a team or an idea and funding them to find their cofounder. Jack Owen, Talent Associate at EF, talks through some of the most common mistakes individuals make when building AI companies.
About Jack
LinkedIn: https://www.linkedin.com/in/jackwowen/