`

Deep Learning

Google released TensorFlow in November 2015 and I started using TensorFlow in the last week of December 2015. That got me really excited as I just started working on some Computer Vision problems. The first problem I did was about Where's Wally?, and it definitely won't be the last project that I will be doing on Deep Learning. Check out the list of questions and answers that I have posted on StackOverFlow! I don't update this regularly now as most of my code is either in an internal repository or uploaded on github. I'm currently in the process of porting all my internal code to a new project that I'm excited to work on.

NYC Taxi Challenge

TLDR The Challenge Understanding the Data Synthesized Understanding of Data Exploring the Data in Pandas Feature Engineering Duration Speed Tip Day of Week Weekend Building the Model Test Dataset Classifier Experiment 1 Experiment 2 Experiment 3 Experiment 4 Regressor Deploying in a production environment? Conclusion TLDR I normalized the data...

Research Assistant

EDIT on 29th December 2017. Deep Learning research is moving so quickly. The idea I had for the paper I was trying to write was released by NVIDIA on December 4th 2017. They probably submitted it for CVPR 2018. Note that I’m not saying I could have achieved their results...

ICML 2017

This is the first ever academic conference that I’m attending. I paid for it out of my own pocket because I wanted to experience first hand what an academic conference is like. This is an uncurated rant. Day 1 - Tutorials Interpretable Machine Learning Recent Advances in Stochastic Convex and...

Relaxation of Optimization and Backpropagation

I spent April 2017 reading up on methods of parallelizing deep learning and looking at distributed methods. I’ve learnt a lot more about deep learning in the past month. The key insight that I got: You don’t have to strictly adhere to backpropagation or optimization fundamentals for Deep Learning to...

Reconstruction

I’ve been exploring TensorFlow lately. I need to get better at it, as I think TensorFlow will have lots of features in future. Here’s an unsupervised autoencoder. I only trained it on a single example so it’s definitely overfitting. I just did it to prove that some code that I’m...

Entrepreneur First

EDIT: This post has been elaborated on. I will walk you through the entire process of my EFSG interviews. It might change in future. I’ll be writing my opinion of each stage, since I think it’s what most people will be looking out for. STAGE ONE The first part of...

Learning from rotations?

Part 1: 4th June 2016 For almost the longest time ever, I had this hypothesis: Take a video from all perspectives of an object, do some segmentation to extract only the object, train the neural network with these images, and you will get a neural network that can recognize these...

Wally Part 2

I was reminded of this task as I had to do a write up on this. As such, I went back to fix the data sets! Unfortunately, I really wanted to put a collage that looks like this to show you the information. SVHN, MNIST, Noise However, this was hardcoded...

tf.image.rotate

Part 1 In my previous post on SVHN TensorFlow, I mentioned that a rotate function would be good! Turns out that someone else also had the same idea too! The issue is now on TensorFlow GitHub. I would like to take this small project on. It will definitely give me...

SVHN TensorFlow

Part 1 In my previous post, Wally TensorFlow, I adapted the code from the MNIST convolutional version. I decided to try out the CIFAR10 one now. This time, I ran it on an openly available data set, which is the Street View House Numbers. The intuition for me was to...

Wally TensorFlow

Before diving into Wally, I tried out the basic TensorFlow tutorials I haven’t gone through all the tutorials yet, but did those most relevant to Computer Vision. MNIST for ML Beginners Deep MNIST for Experts TensorFlow Mechanics 101 Convolutional Neural Networks I will complete the others in time to come,...

Let The Tensors Flow (A Tutorial)

I wanted to implement a neural network myself and see how it would work in finding Wally. Since Google released TensorFlow recently, and it seemed like a really good toolbox, I set out to install and use it. It’s worth mentioning that TensorFlow is moving extremely quickly, so stuff on...

Conventional Image Processing Methods

I was first introduced to Computer Vision in October 2015. I found it really fascinating. The first project that I did with respect to Computer Vision was: Where’s Wally? Where's Wally? I was new to this field and had to do quite a bit of reading to figure out how...