Deployment of smart grids gives space to an occurrence of new methods of machine learning and data analysis. Smart grids can contain of millions of smart meters, which produce a large amount of data of electricity consumption (long time series). In addition to time series of electricity consumption, we can have extra information about the consumer like ZIP code, type of consumer (consumer vs. prosumer) and so on. These data can be used to support intelligent grid control, make an accurate forecast or to detect anomalies. In this blog post, I will focus on the exploration of available open smart meter data and on the creation of a simple forecast model, which uses similar day approach (will be drawn up in detail below).

I address this post to my peers - to my liberal, driven, University-educated and multi-cultural counterparts.

Data science is the buzzword du jour in many circles. While most who are interested in the field have an inkling of its importance, to outsiders it may be difficult to pinpoint exactly what data science entails. Seeing the need for more clarity, Metis Chicago coordinated an event focused on demystifying data science for those intrigued by the field, or perhaps who are interested in breaking into it themselves.

** Sun 06 November 2016

In this post we’re going to do a bunch of cool things following up on the last post introducing implicit matrix factorization. We’re going to explore Learning to Rank, a different method for implicit matrix factorization, and then use the library LightFM to incorporate side information into our recommender. Next, we’ll use scikit-optimize to be smarter than grid search for cross validating hyperparameters. Lastly, we’ll see that we can move beyond simple user-to-item and item-to-item recommendations now that we have side information embedded in the same space as our users and items. Let’s go!

We’ve learned how Artificial Neural Networks (ANN) can be used to recognize handwritten digits in a previous post. In the current post, we discuss additional techniques to improve the accuracy of neural networks. Neural networks have been used successfully to solve problems such as image/audio recognition and language processing (see Figure 1).

One of the most important fields in robotics is the optimization of controllers. Currently, robots are often treated as a black box in this optimization process, which is the reason why derivative-free optimization methods such as evolutionary algorithms or reinforcement learning are omnipresent. When gradient-based methods are used, models are kept small or rely on finite difference approximations for the Jacobian. This method quickly grows expensive with increasing numbers of parameters, such as found in deep learning. We propose an implementation of a modern physics engine, which can differentiate control parameters. This engine is implemented for both CPU and GPU. Firstly, this paper shows how such an engine speeds up the optimization process, even for small problems. Furthermore, it explains why this is an alternative approach to deep Q-learning, for using deep learning in robotics. Finally, we argue that this is a big step for deep learning in robotics, as it opens up new possibilities to optimize robots, both in hardware and software.

This project was done with Yika Luo and Shashank Bhargava, who are also both students at UC Berkeley.