Aside from supervised and unsupervised learning, Reinforcement Learning is the third main domain of machine learning. Popularized notably thanks to its recent achievements in Go, this burgeoning field still seems to have far more untapped potential.

This article presents the architecture of this particular type of machine learning algorithm and some common pitfalls to avoid that are specific to Reinforcement Learning.



At the scale of a town district, a community solar self-consumption project is not only possible technically and legally, but also financially viable today.

In this article, we will discuss ‘allocation keys’ to break down the energy produced between the various participating consumers in order to obtain the desired balance between redistribution of benefits to those who truly consume solar energy and profitability for the investor.

Unsurprisingly, the more expensive the solar energy, the lower the savings. If the price of solar is higher than the average purchase price from the grid, i.e. 8.64ct € / kWh, there is no savings generated at the district level.


Cellular grids bring together prosumers interested in participating in a connected community (including eco-suburbs) and pave the way for peer-to-peer electricity transactions.

In this article, we will discuss how to integrate over time the potentials of the multi-faceted sources of flexibility existing within a community, while promoting local and shared economic models.

Réseaux énergétiques cellulaires peer-to-peer trading


Energy storage is a hot topic these days. Stationary batteries are being added to buildings, microgrids, and combined with solar PV systems for instance.

In this article, we will discuss whether they are economically viable today and explore different energy management strategies to optimize performance and the return on investment.

energy storage, the price of the revolution


What is ‘information’ and what is its relationship to the ‘data’? Both terms are frequently confused and understanding the difference is key for any AI applications.

In this article, we revisit those basic concepts, which touch upon the very essence of BeeBryte’s expertise in developing and implementing its energy intelligence & automation solutions.

Data vs information livre blanc. Les termes "Donnée" et "Information" prête souvent à confusion. Bien que liés, ceux-ci couvrent des notions différentes.


Blockchain is today in the spotlight of many discussions, articles, interviews and books. Deeply related to blockchain, and basically at its heart, the vast and fascinating field of cryptography is used everywhere, and has been an integral part of BeeBryte’s energy intelligence & automation solution since its inception.

This article intends to give you a first flavor of this science of secrecy underpinning our secured systems, and paving the way to the buzzy Blockchain.

Cryptography and blockchain


Machine Learning is a particular field from the wide range of solutions grouped under the generic name of artificial intelligence, as well as expert systems and many optimization techniques. It is usually subdivided into three subdomains: Supervised Learning, Unsupervised Learning, and Reinforcement Learning.

In this article, we will focus on the study of Supervised Learning, which consists in designing a mathematical autonomous model.

Machine learningis a piece of cake