Posts
-
Implementing a Simple Neural Network from Scratch
Artificial intelligence pervades our daily lives, from ChatGPT to mobile phone handwriting recognition. Even without delving into the mathematical intricacies, frameworks like PyTorch and TensorFlow simplify neural network implementation. Yet, understanding the mechanics behind these networks enhances our grasp of such frameworks. In this post, we’ll construct a basic feed-forward neural network to classify handwritten digits from the MNIST dataset, a repository of images spanning...
-
Creating a Generalized Additive Model from Scratch
When building a good statistical model, we all know that there are many options in the statistician’s toolkit. In my line of work, I need to be able to quickly compute and interpret the results of models, so linear and logistic regression methods are my best friends. Although these models are easy to interpret, they aren’t as flexible or powerful as other options. For example,...