Why we should all think more carefully about what we hear and say I bet you didn't wake up today expecting to be advised to reinvent the wheel. These days, the internet is chock full of articles containing well-meaning tips directed toward aspiring programmers.
3 Questions I was Afraid to Ask (and my Tensorflow 2.0 Template) For many people, myself included, Q-learning serves as an introduction to the world of reinforcement learning. It gets us neatly accustomed to the core ideas of states, actions and rewards in a way that is intuitive and not bogged down by complicated technical details.
Empirically Examining their Equivalence Bayesian methods of performing machine learning offer several advantages over their counterparts, notably the ability to estimate uncertainty and the option to encode contextual knowledge as prior distributions. Why then, aren't they more widely used?
It processes screen and minimap feature layers with... two layers with 16, 32 filters of size 8, 4 and stride 4, 2 respectively. The non-spatial features vector is processed by a linear layer with a tanh non-linearity. The results are concatenated and sent through a linear layer with a ReLU activation.
Deconstructing the Art Raise your hand if you've heard these before: "Hyperparameter tuning is more of an art than a science." "Tuning Machine Learning models is like black magic." These are nice, evocative statements that may serve well in instilling a sense of intrigue towards our work as Data Scientists in the laypeople we have casual conversations with.
Functions and Derivatives Ideas from mathematics underlie virtually every technique and concept in Data Science. While possessing a rigorous understanding of all these ideas is certainly not required, it can be beneficial.
Standardization vs. Normalization Data Science is an exciting field that combines many disciplines and has brought academia and industry together in spectacular fashion. For the beginner practitioner, there are a wealth of resources available for them to learn from, and career opportunities to match.
Define some objective, then let a population of algorithms compete against one another, mixing and refining characteristics of the most successful of their predecessors, and occasionally introducing novel approaches. I'm speaking, of course, of machine learning competitions. They are rather ubiquitous in today's research and commercial environment, but what are they good for, and what should one know about them?
Visualizing results can be a powerful form of motivation and preparation. However, in the fitness domain, it can often be difficult to clearly see this future outcome. Can we use deep learning to bring people closer to their individual fitness goals by helping them visualize their future results?