I am a machine learning researcher at Camb.AI. I post about deep learning, electronics, and other things I find interesting.
Sign up for email notifications of new posts: mbaas.substack.com
A fully labelled time series dataset of the vote counts and associated data from the South Africa 2024 general election, along with some preliminary analysis.
The design and construction of a portable biltong dryer.
My particularly uninformed opinion and speculation on when and if we might get to use actual human-level AI.
A design and implementation of a minimalist 3D printed case for a Prusa i3 MK3S+ 3D printer.
A short practical introduction to the hierarchical global style token (HGST) layer occasionally used in speech processing tasks.
A brief summary of the trends in Eskom loadshedding over the past 5 years.
An abridged taxonomy of the major bitcoin address formats and versions.
A major update to my simple-speaker-embeddings repository with vastly improved training and new pretrained model.
Making a functional air quality sensor with a raspberry pi and a few sensors, with a companion app.
More statistical analysis of Steam review trends to uncover some more hidden heuristics.
A few guidelines for judging a game’s Steam review score at a glance, by treating Steam reviews as random variables and processes.
A quick guide to our second place solution for the Zindi UmojaHack South AFrica Yassir ETA prediction challenge.
Using UNets to estimate H-plane electric field magnitudes of arbitrary dipole antenna arrays in real time.
A quick intro to simulating the electric field for simple current distributions in python using MEEP.
A quick rundown of the basic antenna parameters, in the notation used at Stellenbosch University.
Think Gmail smart compose, but for technical LaTeX writing on Overleaf.
Building a simple charge pump with only resistors, diodes, npn BJT’s and capacitors.
Pretrained PyTorch Resnet models for anime images using the Danbooru2018 dataset.
Performing simple X-ray diffraction (XRD) classification with convolutional neural networks.
A quick simple project to make a simple motion detecting camera with the minimum requirements. The end result is to have a Raspberry Pi that sends you a short video on Telegram when it detects motion.
A quick post connecting the ideas of the convolution operation in mathematics & signals analysis to deep learning convolution.
This aims to explain my understanding of what people refer to when they talk bias and variance in the context of a function approximator.