Using the PennyLane Quantum Machine Learning library it’s easy to create a quantum/digital-hybrid equivalent of the glorious LSTM layer. — In recent years, Toronto-based startup company Xanadu has introduced a python framework called PennyLane which allows users to create hybrid Quantum Machine Learning (QML) models. While it’s still too early to claim that quantum computing has taken over, there are some areas where it can give an advantage as for…

Don’t know how to set up your first project? Begin your adventure in Machine Learning by adapting this skeleton code to your needs. Voilà! — This post is a walkthough of a Google Colab notebook you can access from this link. To begin with, we generate synthetic inputs for a binary classification task with many input features. See the scikit-learn documentation for further information. Not all input features are informative. A standard technique called…

The infamous 60 Hz cycle hum is an intrinsic problem of single coil electric guitars. Among the many alternatives to tame unwanted noise, one stands out as the geekiest and more effective. This is how I made it without breaking the bank. — Recently, I have purchased a previously-owned Fender Stratocaster Made in Japan. It’s an ok guitar for both playing and tinkering with it. All along, what I had in my mind was to buy an instrument that was able to upgrade or mod. The stock pickups were ok, but I decided…

While buying a kit is the shortest path, is it really worth it? A long and winding road taught me a lot about how many things can go wrong and how to fix them. — I have been playing electric guitars since I was a teenager. Since I owned a few, I believed I had a decent understanding of what these instruments actually were, but I was wrong. As COVID-19 dragged on, I decided to start a new hobby. As described in a previous post…

We’ll see how to use an op-amp and clipping diodes to distort the signal from an electric guitar. — When it comes to distortion, almost every electric guitar player has faced the dilemma of whether to use a transistor-based pedal or to turn all the way up the gain of a valve amplifier. It is common lore that while the latter sounds better, the former is usually cheaper and…

The geekiest thing I’ve done. And I’ve learned so much in the process. — When I was 17, my parents bought me an electric guitar. As a matter of fact, it was my guitar teacher (“maestro” Antonio Scarpetti) who made the purchase because he thought I didn’t know enough at the time. It turned out to be an affordable Ibanez RG270DX, Made in Korea…

A combination of statistical methods and engineering is key. Let’s see why and how KBinsDiscretizer should be the ML engineer’s friend. — Data analysis can be seen as a device (or a black box; or a Chinese room) which, given some inputs, provides outputs. The main difference between rule-based and statistics-based (aka machine learning) is that the former accepts training examples and rules (“algorithms”) as inputs and gives predictions as output, while…

If ignorance is bliss, knowing how to deal with incomplete knowledge is the key. Bayes’ Theorem formalizes all of that. — Bayes’ Theorem formalizes how to calculate the probability of an event to occur given our incomplete knowledge about other factors affecting its realization, and our pre-existing belief. For example, is it going to rain today if the sky is cloudy? Let’s start with some basic definitions. The probability of an…

It’s easy to be confused and not see the (random) forest for the trees. Let’s take a closer look at some of the most common methods in Machine Learning to analyze heterogeneous data for both classification and regression tasks. — Decision Trees are a very common, intuitive and explainable method of analyzing data with the goal of dividing the entities into separate groups (classification) or predicting a quantity (regression). DT’s are also great at handing heterogeneous un-normalized input features (e.g. both categorical and continuous).