a blog about what's cool to me
I’ve been spending lots of time looking at names recently. The US Government posts a dataset of the yearly popularity of baby names from 1880. I wanted a nicer way to let the names tell me about the relationships between them. I therefore made a 3d name graph to let name historical data speak for […]
Making a name graph from their popularity over time Read More »
It’s hard to decide on a baby name, even more so when multiple people are invested in the decision-making process. My friends recently had a baby and couldn’t decide between Alex and Niklas. I naturally took it upon myself to solve their problem by creating a Machine Learning algorithm that finds the names laying perfectly
Using Machine Learning to Compromise between names Read More »
I study the effect of dataset poisoning on popular datasets: MNIST, fashion-MNIST, CIFAR10, CIFAR100 to study its effect.
The effect of Dataset Poisoning on Model Accuracy Read More »
We often use recurrent neural network architectures for use cases such as streaming, where we make a new prediction each time new information arrives. It’s common to give it a long context each time, which repeats work and is therefore computationally wasteful. In this blog post, I will illustrate multiple ways to perform inference step
Step by step recurrent neural network inference with Keras Read More »
Jupyter notebook here I love playing Beat Saber, a VR rhythm game in which you slice blocks in time with music. Turning a song into a Beat Saber map requires serious time, dedication and talent. I’ll attempt to bypass the requirements by trying to use artificial intelligence to make the Beat Saber maps. In order
make beat saber maps using machine learning (1) Read More »
I love flying around in flight simulators. As fun as it is to fly around and try to figure out procedures, it’s rarely a game as such. There are no missions, no goals, no penalty for mistakes. FSEconomy is a plugin for flight simulators which adds a layer of gamification by enabling users, among others,
Plotting tool for FSEconomy Read More »
In the past, I’ve covered how to implement k-means and how to cluster MNIST directly. Then I showed a bad way to reduce dimensionality for clustering. Finally, we used an autoencoder on MNIST. In this post, we’ll join everything together by performing clustering on the encoding and see how it performs. In the previous post,
Clustering on encoding Read More »
Dimensionality reduction is useful in compressing the representation of your data. In my last post, how not to reduce dimensionality for clustering, I show a neural network that reduces the 784 dimensions of the MNIST dataset into 3, which are constrained by the k-means algorithm to capture the essence of the data. This week we’ll
Reduce MNIST dimensionality using autoencoders Read More »
In my previous post, I followed up on my k-means tutorial by applying it to cluster MNIST. We’re used to reading digits, so the centroids made perfect sense to us since plotting them just looked like a smudged number. The dataset even clustered pretty well since all digits are centred and look similar enough that
How not to reduce dimensionality for clustering Read More »
This is the beginning of a series in which we’ll explore ways to perform clustering. In particular, we’ll perform clustering on the MNIST dataset, since it’s a dataset of high dimensionality which is easy to understand. In my previous post, ‘How to implement the K-means algorithm in Tensorflow‘, I explain the code of Sergey Kovalev,
Clustering MNIST – naïve Read More »