Thanks for visiting this page about the Collections project.
You may have hovered over the tab, finding many links being under development for now. This page serves as an explanation for this observation, as well as a rationale for the Collections project.
Because what is this Collections project?
Please let me explain.
On this website, I write many types of articles. Many of them revolve around one particular topic and thus cover a hyper niche. Examples of this type of articles is my blog about chaotic neurons in which I attempted to replicate the results of an Arxiv paper – only to find a fellow engineer argue that the model is not better than vector based averaging. I love those kind of posts.
However, if you’re really keen on learning stuff, you must do deliberately. That is, you must practice every day for some time, preferably one hour or so. In my view, you must additionally build on top of your existing knowledge, and relate your findings to the more general concepts. That’s what I’m trying to do with the Collections I introduced slightly earlier.
For example, suppose that you’re interested in learning Keras, one of the important Python frameworks for deep learning that are used today. If I write a blog every now and then about Keras, you may pick up some clues here and there, but you will never develop a full understanding of the framework and its power. And neither will I. Despite the fact that I have been fortunate to develop a lot of knowledge over the past year, it’s never time to stop learning new things, expanding your knowledge on the go.
So far, these will be the Collections I will be working on over the next year – starting in the summer of 2019:
- Mastering Matplotlib, in which I will attempt to get to the bottom of this Python framework for visualizing data sets, explore its two widely-used APIs.
- Getting to know Keras much deeper than I do now. Although I use Keras on a daily basis, there’s probably a lot more that I’ll need to learn before I can say that I can dream Keras code… creating a Collection is a nice way of trying to achieve just that 🙂
- The same goes for Pytorch, with which I have substantially less experience. I personally think this is bad, since it’s also one of the most widely used frameworks used today. Hence, will start working on a Collection 🙂
- Scikit-learn is one of my favorites for training traditional machine learning models like Support Vector Machines and so on.
- Finally, if you wish to understand the framework deeply, you’ll also have to understand the concepts deeply. I will by consequence of dissecting the framworks also focus on the underlying concepts, being the components of neural networks, the methods for optimizing them, and so on. I will attempt to do the same for the traditional methods, but this will be started only slightly later.
I’m very much looking forward to seeing these Collections grow, and I hope that you’ll bookmark this page (or this website) to find new blogs on the topics in the near future. Please, do not hesitate to get in touch if you wish to ask questions, send remarks or think I’ve made a mistake in one of my blogs. I’m happy to answer, to discuss and to correct! 🙂 Please write your comment(s) below the respective blogs, in the comments field. As they need to be accepted by me in order to fight spam, it may take a couple of days for them to become visible.
Happy engineering! 🙂