- This event has passed.
Data Science Meet-Up
March 6, 2019 @ 6:30 pm - 8:30 pm MST
FreeStartup Edmonton
Wednesday, March 6 at 6:30 PM
• “Cost-Sensitive Classifier Evaluation Using Cost Curves” – Professor Robert Holte (CS @ UofA)
• “Figuring out the limits of human vision with deep learning” – Navaneeth Kamballur Kottayil (Machine Learning Researcher at BorealisAI)
Pizza will be provided before presentations and after presentations we will head to a pub for networking!
The meetup is sponsored by AltaML (http://altaml.com/).
Abstracts:
Cost-Sensitive Classifier Evaluation Using Cost Curves
(co-author: Chris Drummond)
The evaluation of classifier performance in a cost-sensitive setting is straightforward if the operating conditions (misclassification costs and class distributions) are fixed and known. When this is not the case, evaluation requires a method of visualizing classifier performance across the full range of possible operating conditions. This talk outlines the most important requirements for cost-sensitive classifier evaluation, and introduces a technique for classifier performance visualization – the cost curve – that meets all these requirements. This talk should be of interest to anyone who works in areas that use classifiers, for example, machine learning, pattern recognition, biometrics, and diagnosis.
Related papers:
http://dx.doi.org/10.1007/s10994-006-8199-5
https://webdocs.cs.ualberta.ca/~holte/Publications/kdd2000.pdf
Figuring out the limits of human vision with deep learning
In this talk, I present a part of my PhD thesis done in collaboration with Telecom Paristech. In this work, we explored methods to derive the human perceptual limit to see changes in an image. We proposed a new method to solve the problem with convolutional neural network. Our solution involved using deep learning in conjunction with traditional methods in psychophysics. Our method derives the human error detection threshold by solving an alternative problem of image quality assessment. We tested our method on publicly available databases for measuring human visual thresholds and show state of the art results.
Related papers:
https://arxiv.org/abs/1712.07269
https://www.researchgate.net/publication/327994999_Learning_Local_Distortion_Visibility_from_Image_Quality
Edmonton Data Science meetup is a place to learn about data science and network with fellow Data Scientists. The presentations for the upcoming meetup…
