06 Dec 2019 · 18 mins read
An introduction to Generative Adversarial Networks (in Swift for TensorFlow)
Generative adversarial networks, or GANS, are one of the most interesting ideas in deep learning. Using GANs computers get a sense of imagination, they can create their own “things”. But how do they do that? It’s easy to have a computer generate random data, but that’s of low value to...
04 Dec 2019 · 1 mins read
An Empirical Comparison of Optimizers for Machine Learning Models
At every point in time during training, a neural network has a certain loss, or error, calculated using a cost function (also referred to as a loss function). This function indicates how ‘wrong’ the network (parameters) is based on the training or validation data. Optimally, the loss would be as...
03 Dec 2019 · 16 mins read
Your first Swift for TensorFlow model
Swift for TensorFlow is TensorFlow, implemented in Swift. Its Python counterpart (“TensorFlow”) is one of the best deep learning libraries ever made, but it’s hitting its limits. Python, being interpreted, is quite a slow language. Python is also limiting in many other ways, it lacks features that could make your...
14 Nov 2019 · 1 min read
Benchmarking deep learning activation functions on MNIST
Over the years, many activation functions have been introduced by various machine learning researchers. And with so many different activation functions to choose from, aspiring machine learning practitioners might not be able to see the forest for the trees. Although this range of options allows practitioners to train more accurate...
13 Oct 2019 · 1 min read
Building a Barcode Scanner in Swift
Barcodes are everywhere. They provide a uniform interface for machines to identify real world items. Thanks to the decimal code underneath the bars, barcodes are also interpretable by humans, which makes them wildly used virtually everywhere.
14 Aug 2019 · 4 mins read
Lesson 14 - An end at the beginning
For the final lesson of fast.ai I redid the first project of the course. In Swift. (Read the first post so you know what we’re building) After using Swift and TensorFlow for the first time yesterday (post), I want to use Swift for everything. So I figured it might be...
13 Aug 2019 · 7 mins read
Lesson 13 - Linear Regression in Swift with Tensorflow
Swift is a new programming language introduced by Apple in 2014. Despite being so young compared to other languages, it is already widely used in industry. It is mostly used to develop apps for Apple’s platforms. Recently however, after Apple making Swift open source, Swift was ported to linux and...
12 Aug 2019 · 5 mins read
Lesson 12 - Label Smoothing
Working in machine learning includes dealing with poorly labelled datasets. Very few companies can afford hiring people to label the huge amounts of data required for large scale projects. Luckily, high quality datasets are available for practise projects. In production however, one will most likely need a custom dataset. With...
11 Aug 2019 · 4 mins read
Lesson 11 - Adam Optimzer in Python
Optimizers are one of the most important components of neural networks. They control the learning rate and update the weights and biases of neural networks using the gradients computed by backpropagation. There are many different types of optimizers. Stochastic gradient descent, or SGD, and Adam are by far the most...
10 Aug 2019 · 2 mins read
Lesson 10 - Correlation vs Covariance
Machine learning is for a very large part about the relations between numbers. This problem, however, is not limited just to machine learning. It is a problem in mathematics, statistics more precisely, as well. In this post I want to give you a quick introduction to statistics by explaining the...
09 Aug 2019 · 15 mins read
Lesson 9 - Looking inside a neural network
While sometimes neural networks seem like black boxes that happen to be good at certain tasks, they really are not. After a quick setup of a fully connected neural network with one hidden layer, I’ll demonstrate how to look inside a neural network by visualising the hidden layer and get...
08 Aug 2019 · 6 mins read
Lesson 8 - Numpy vs PyTorch for Linear Algebra
Numpy is one of the most popular linear algebra libraries right now. There’s also PyTorch - an open source deep learning framework developed by Facebook Research. While the latter is best known for its machine learning capabilities, it can also be used for linear algebra, just like Numpy.
07 Aug 2019 · 3 mins read
Fast.ai Part 1 - Conclusion
For the past week I did one lesson of fast.ai a day including a machine learning project for each lesson. I kept track of the things I learned on my blog. Here’s a summary of what I learned so far.
07 Aug 2019 · 10 mins read
Lesson 7 - GANs for Watermark Removal
Generative Adversarial Networks, or GANs, are a new machine learning technique developed by Goodfellow et al. (2014). GANs are generally known as networks that generate new things like images, videos, text, music or nealry any other form of media. This is not the only application of GANs, however. GANs can...
07 Aug 2019 · 6 mins read
Lesson 6 - Convolutions and CNNs
Convolutions are ways to transform images. There are two main applications of this technique: image editing and convolutional neural networks. Convolutional neural networks, or CNNs, use convolutions to transform the image making them perform much better at computer vision tasks.
06 Aug 2019 · 7 mins read
Lesson 5 - Neural Network Fundamentals
To really understand deep learning, you need to understand how neural networks, the mechanism that makes deep learning deep, work. This tutorial explains how to write a neural network from scratch using PyTorch.
05 Aug 2019 · 5 mins read
Lesson 4 - Autocompletion while typing
Virtually all mobile phones support autocompletion nowadays. This might look like a casual feature but when you start thinking about how to implement this, something I’ve been wanting to know for a very long time, it can be quite complex. If you know a little bit about deep learning, it’s...
04 Aug 2019 · 9 mins read
Lesson 3 - Self Driving Cars
Deep learning enables us to do things that could never be done before. It is able to do things that were science fiction only 20 or even 10 years ago. One of those things is having cars drive themselves. If I were old enough to drive a car, I would...
03 Aug 2019 · 1 mins read
Lesson 2 part 3 - Common Issues with Neural Networks
This post outlines 4 common issues that cause your neural network to perform poorly along with symptoms and explanations for the behaviour.
03 Aug 2019 · 8 mins read
Lesson 2 part 2 - Gradient Descent from Scratch in Pytorch
Understanding gradient descent is a critical part of learning machine learning. Although gradient descent is not the preferred method for optimization, it is similar to the more advanced methods so it’s still valuable to learn how it works.
03 Aug 2019 · 10 mins read
Lesson 2 Part 1 - ML in Production
After my blog post on the first lesson of fast.ai yesterday (which got recognised by the founder of fast.ai!), today is time for the second lesson. Because this lesson consisted of three distinct parts, I decided to write 3 separate blog posts. The first one, the one you’re reading now...
02 Aug 2019 · 5 mins read
Lesson 1 - Training a fruit classifier
After finishing CS229 Machine Learning by Stanford yesterday I decided to continue this fascinating journey with the fast.ai course. And wow! The first impressions are extremely good.
28 Jul 2019 · 7 mins read
Introduction to natural language processing in Swift
As computers get smarter, the communication between machines and humans becomes more of a bottleneck. While humans are socially smarter, computers have surpassed us in many ways in areas like math and science. Perhaps the most important side of this bottleneck is communicating emotions. Although emotions are a fundamental part...
23 Jul 2019 · 9 mins read
Compressing images using Python
Compressing images is a neat way to shrink the size of an image while maintaining the resolution. In this tutorial we’re building an image compressor using Python, Numpy and Pillow. We’ll be using machine learning, the unsupervised K-means algorithm to be precise.
10 Jul 2019 · 6 mins read
For my WWDC Scholarship submission I used a custom machine learning model that I trained using CreateML. In this blogpost I will be explaining how I went from binary data to a state of the art machine learning model.
10 Jul 2019 · 13 mins read
On April 16th, 2019 an email landed in my inbox. The subject line read: “You’ve been awarded a WWDC19 Scholarship.” What?! Because this was my first time taking part in a programming competition, I didn’t expect to win at all.