You can't miss this Thursday evening program: 1) A hot Zurich based startup demonstrating computer vision for automatic video analysis, 2) A brand new study of text sentiment analysis tools. AND - wait for it - 3) A very useful tutorial in python on how to use the popular open source machine learning library, scikit-learn, given by one of its core developers! (video conference)
Abstracts for meetup #2:
• webcamaze - Amazing Webcams
Co-founder upicto and webcamaze
Webcams simply overwrite each image with the next incoming one. This is a pity! With webcamaze we build a system that automatically analyses more than[masked] webcams all around the world. It summarizes the continuous image streams, spots interesting images and puts them aside for people to enjoy. An obvious challenge is how to cope with the enormous amount of data generated. This calls for light-weighted methods which run in an online manner. A second, even more challenging task from the computer vision perspective, is to determine what humans consider as "interesting" and how this might be implemented computationally. In this talk I will show some insights of our approach and discuss recent results and limitations.
• #like or #fail - How Can Computers Tell the Difference?
Associate Professor at Zurich University of Applied Sciences (ZHAW) Datalab
Sentiment analysis appears to be one of the easier tasks in the realm of text analytics: given a text like a tweet or product review, decide whether it contains positive or negative opinion. This task is almost trivial for humans, but it turns out to be a true challenge for automated systems.
In this talk, we will explain the intrinsic difficulties of automated sentiment analysis; describe existing solution approaches; analyze performance of state-of-the-art sentiment analysis tools; and show how to improve their analysis accuracy using machine learning technologies.
(short break until about 20:00)
• Machine learning and data exploration in Python with scikit-learn (video conference)
Machine learning researcher at Amazon, and core developer at scikit-learn
Data exploration and fast prototyping are important aspects of machine learning. The python ecosystem provides a lot of tools to help with these tasks, and incredible interactivity. The talk will give a quick walk through of how to use scikit-learn to quickly develop working systems and get a feel for your data. We will discuss model selection and the creation of full pipelines.
Also, here are the videos from our first meetup in February, in case you missed it. Please share!