Thoughts on data science, statistics and machine learning.
Weighted Loss Functions for Instance Segmentation
This post is a follow up to my talk, Practical Image Classification & Object Detection at PyData Delhi 2018. You can watch the talk here: and see the slides here. I spoke at length about the different kinds of problems in computer vision and how they are interpreted in deep learning architectures. I spent a fair bit of time on instance and semantic segmentation (for an introduction to these problems, watch Justin Johnson’s lecture from the Stanford CS231 course here).
How Not to Ask Questions After a Talk
Dedicated to the memory of Mrutyunjay Mishra (M2). He was a torrent of ideas. He encouraged me to write this post because as much as he talked endlessly without pausing to breathe, he hated wasteful discourse. New Delhi, January 2024 PyCon India 2017 ended last weekend in Delhi. The conference escaped the infamous winter smog-storm by a whisker. In the last few years, PyCon India has grown to become the largest PyCon outside of North America, with over a thousand participants attending from all over the country.
Playing With the Konmari Method
I heard about the bestseller The Life-Changing Magic of Tidying Up at a SciPy talk about deculttering your data science project. The speakers admitted they hadn’t read it - they were simply trying to point out that tidying up your space and tidying up your software project are both similar. I’ve been married and living with my wife for about a year now. After we moved into “our own home” last year, we have both undergone major role reversals when it comes to tidying up.
On the Linearity of Bayesian Classifiers
In his book, Neural Networks - A Comprehensive Foundation, Simon Haykin has an entire section (3.10) dedicated to how perceptrons and Bayesian classifiers are closely related when operating in a Gaussian environment. However, it is not until the end of the section that Haykin mentions that the relation is only limited to linearity. What is interesting about this is that a Perceptron can produce the same classification “model” as a Bayesian classifier, provided that the underlying data is drawn from a Gaussian distribution.