In this article I’m going to go through how to set up CNTK with Visual Studio Code and take advantage of those PASCAL GPUs I know everybody has these days. I will also do a breif overview of what CNTK and Visual Studio Code are and why they are so incredible for machine learning scientists.
So today we are going to do something really awesome. Operationalize Keras with Azure Machine Learning. Why in the world would we want to do this? Well we can configure Deep Neural Nets and train them on GPU. In fact, in this article, we will train a 2 depth neural network which outputs a linear prediction of energy efficiency of buildings and then operationalize that GPU trained network on Azure Machine Learning for production API usage.
So I just completed an incredible project with Brain Thermal Tunnel Genix, where I learned so much about pattern recognition, machine learning and taking research and algorithms and pushing those into a production environment where it can be integrated into a real product. Today’s article takes those lessons and provides a sample on how to perform complex modelling and operationalize it in the cloud. The accompanying Gallery Example can be found here.
So there are a ton of articles out there on the theory of Reinforcement Learning, but very few with an actual application. I watched a few lectures from Berkley, and read a few articles by NVidia and thought, “Well, lets just give this a shot”. 8 hours later, this is what I had.
Herby V1 simply learns to go forward as much as possible while avoiding obstacles.
So here is a pretty raw blog article; not unlike most of my articles. The cognitive revolution. I’m going to coin this term today. What the heck is this thing? What does it mean for you? What does it mean for me? Where did it come from? What is it? These are questions I aim to answer in this blog article.
So this blog post is to get you operational with Docker, image and volume management with a pivot towards scientific computing and tensor flow. So I am working on building a Jupyter Notebook for the local mahcine learning meetup to learn the ins and outs of Tensor Flow and deploy this thing up to Azure. Part of getting this to work is not only managing the Docker Containers, but also the data on the volumes so when we deploy up to Azure and somebody opens up the notebook it comes pre-loaded with all the necessary tutorial data.
So this is an interesting problem. You are collecting data from somewhere and you want to feed it into a neural network for classification. There is one main problem with this. The shape of the data! Neural networks and really just anything require specifically shaped data, you can’t just like give it something of ambiguous size. There are tons of papers out there on dimensionality reduction, but nothing on dimensionality reduction to a specified size. This article explains my approach.
This article is loosely based on a time series challenge from customer data. I have fabricated 3 data files such that they represent the same challenge and we will go through the process of discovering that data. The primary challenge in this data set is that it is from a sleep study and the researchers left the date portion of the time stamp off. What this means is that at midnight, the data plots at the beginning of the x-axis. The second challenge is lining up data to see if there is anything interesting with the time. So yes, you can simply plot using the index that python generates, however I’m also interested in the actual time itself as this is a study involving humans.
So today, I was asked to put some thought into what we should focus our entry level data scientists on in terms of tech skills. After I put a bunch of thought into it, I ended up coming up with this. I decided that the most important aspect of this was a few items fold
Don’t overload them
Can deliver to production where the target can be anything, including IoT.
They will not be concerned with building front ends.