Inspirational Open Sourced Projects
The online community for coders is one of the warmest communities out there. The advocation for open-source projects, along with the support which goes with it is jaw dropping, and for that, people who work on these public repo’s are real gems.
Coding standards and requirements have changed so much over the past 20 years. The earlier languages (like C++, HTML) were notoriously difficult and painful to learn but recent languages have made development so much easier and with that, lowering the barrier to entry to great projects.
We see below that some insane tech is now just…open…and free…with no catch. From advanced visual recognition to advanced NLP: it’s nothing short of insane the kind of tech you can get access to for free.
In what follows, I share 4 really cool projects that I absolutely love, and also, are some of the top trending projects on GitHub. Hope you enjoy, and let me know if you have any questions!
Project 1: Sense
Action recognition is an incredibly difficult task to automate because every human is different, so for a camera to be able to generalise and pick up your actions requires a lot of data.
Sense is an engine that uses neural networks to recognise actions. Moreover, they’ve made the model efficient so it’s pretty light on its feet and can easily be deployed. The engines here are trained on millions of videos of humans performing the a variety of actions.
If you want, you can even play with the additional features to calculate count given the action. I mean, it’s pretty cool.
Project 2: Tensortrade
If you know anything about reinforcement learning you know that using it for trading seems logical. However, it’s a bit more complicated then that because state dynamics in Finance change through time, so it’s hard to use the past to predict the future.
However, Tensortrade is still in Beta, but looks to help users to grips with the theory of building, training and deploying automated trading algorithms. It’s meant to have been made pretty extensible, which means that if you have want to include your own features, or, want to incorporate this software into your engine, it should be possible.
The main thing that Tensortrade try to do is to make it easy and fast to test algo strategies but I feel that sometimes, a small barrier may protect a lot of people from using something potentially flawed. Anyways, try it out and make your own opinion!
Project 3: MLFlow
Deploying a big machine learning project is no fun but MLFlow are here to help. The ‘platform’ assists users in packaging machine learning projects into reproducible codes and helps users to share the models as well. A lot of the hard stuff comes included too, like logging, Conda/Docker, also a centralised model store.
For those of us who build and deploy numerous machine learning models (I’m looking at you in the NLP space), then this is good.
Project 4: spaCy
My NLP Brothers!
And what have we got here?
spaCy is awesome. Like it’s really, really good. It essentially provides a lot of the groundwork for state of the art NLP modelling in Python. It’s been built with the latest research so if you’re trying to replicate results from a paper, this is a very good place to start.
Advanced problems can be sucky to start off with but spaCy makes light work of tokenization, parsing, languages, entity recognition and all the other problems we have to think intently about. Because the API removes any complexity from it, you’re able to spend more time focusing on the difficult problems like prediction or inference. Moreover, it’s pretty quick.
spaCy is awesome. No ifs, no buts.
There you have it. I’ve offered 4 of my recent favourite github projects and I hope you enjoy them as much as I do!
Thanks for reading! If you have any questions, please let me know!
Keep up to date with my latest articles here!
Code Python 