Research log
Articles and project write-ups on machine learning, computer vision and systems software.
Machine learning
Activation Functions and their purpose: Binary, Linear, ReLU, Sigmoid, Tanh and Softmax
Binary, linear, sigmoid, softmax, tanh and ReLU activation functions.
Car Detector v2.0: Let's focus on ML
Vehicle extraction and classification with YOLOS, XResNet, FastAPI and a custom inference pipeline.
Car Detector: a month long journey
Car make, model, year and color classification from a 20,000-image dataset.
Hyperparameter Tuning With Bayesian Optimization
Gaussian processes and Bayesian optimization applied to hyperparameter search.
Transfer learning experiment: ResNet with CIFAR-10
Fine-tuning ResNet-50 on CIFAR-10.
Optimization Methods in ML
Feature scaling, mini-batches, batch normalization, momentum, RMSProp and Adam.
Regularization in Machine learning
L1, L2, dropout, data augmentation and early stopping.
Summary of the AlexNet paper: one of the most influential papers in computer vision
Architecture notes from the AlexNet paper.
Time series forecasting: When to invest for Bitcoin
An LSTM experiment on hourly Bitcoin prices, including data exploration and model evaluation.
Systems
C compilation under the hood
C preprocessing, compilation, assembly and linking.
C static libraries and how to use them.
How static archives are built, indexed, linked, and used in C programs.
Hard links vs Symbolic links, What’s the difference?
How inodes and directory entries give hard and symbolic links different behavior.
Static Libraries vs Dynamic Libraries in C
Static and dynamic linking in C.
What Happens When you Type “ls -l *.c” in the shell.
Shell parsing, glob expansion, process creation and program execution.