Microsoft Machine Learning & Data Science Summit is taking place in conjunction with Microsoft Ignite at Georgia World Congress Center.
Today, day 1 started with keynote by Dr. Joseph Sirosh who identified three axes of innovation along with various customer case studies.
Thought leaders and Microsoft engineers discuss the latest Big Data, Machine Learning, Artificial Intelligence, and Open Source techniques and technologies along with important case studies.
There were various great take aways from sessions. For instance esmart systems Davide Roversie talked about challenge of recognizing smaller object, class imbalance issues, mixing real and synthetic images and Crowd flower talked about merging human and computer intelligence. ai.crowdflower.com
Text analytics with LUIS is a great tool to work with, and Dannielle Dean - Sr. Data Scientist at Microsoft gave an overview of various recent advancements in sentiment analysis. We met bots http://ignitebot.azurewebsites.net/, and were told that bots infused with intelligence are the future, which I think is a great take away.
Data Science for Everyone by Brandon Rohrer was an absolute delight. The intro was
New to data science and analytics? Get your foundation here. This talk is a comprehensive overview of how to do data science from start to finish for the brand new data scientist. Brandon Rohrer will introduce you to the ideas and best practices in the field—no math, programming, or previous experience required. You won't leave an expert data scientist, but you'll certainly have a running start.
Code / Slide Deck
Brandon spoke about feature engineering, coefficient of determination, feature enigneering tricks, problem categories in terms of regression, classification, clustering, reinforcement learning,, spectrum of issues, and people who would not believe regardless of what data says
Several side disucssions on Volatility of data/models, correlation vs. causation,and Eat cheese... and risk strangulation, and confirmation bias jokes, but you had to be there.
Brandon said that learning subject matter expertitse and gaining more knowledge about learning domain has always paid off in quality of results, which is a great takeaway.
References
- https://channel9.msdn.com/Events/Machine-Learning-and-Data-Sciences-Conference/Data-Science-Summit-2016
- https://github.com/mwinkle
- https://twitter.com/_brohrer_
- https://www.youtube.com/watch?v=tKa0zDDDaQk&feature=youtu.be
- http://ignitebot.azurewebsites.net/
- http://brohrer.github.io/
- https://github.com/brohrer
- https://github.com/caesar0301/awesome-public-datasets
- https://arxiv.org/abs/1512.03385
- https://lukas.github.io/p1.html
- azure data catalog - https://notebooks.azure.com/library/python4DSv1
