According to research by IDC, applications incorporating advanced and predictive analytics, including machine learning, will grow 65 percent faster than apps without predictive functionality. Ever wondered how Netflix just knows the shows you must watch. And YouTube's recommended playlists are making the job so much easier? Or how Spotify throws up just the perfect songs and Amazon makes you buy stuff even before you search what you want? Machine learning (ML) has transformed the way technology interacts with users.

I love it when I get feedback from a blog that I've written. I appreciate the different perspectives and insights that others bring to a topic of interest. And no blog that I've written has drawn more comments than my blog, "Isaac Asimov: The 4th Law of Robotics." The section of the blog that fueled the most comments stem from a scene in the movie I, Robot where Detective Spooner (played by Will Smith) is explaining to Doctor Calvin (who is responsible for giving robots human-like behaviors) why he distrusts and hates robots. He is describing an incident where his police car crashed into another car and both cars were thrown into a cold and deep river – certain death for all occupants.

MPs from the House of Commons inquiry into fake news were warned last week of a new AI technology that is about to change the world, and not for the better. "We're rapidly moving into an era where the Russians, or any other adversary, can create our public figures saying or doing things that are disgraceful or highly corrosive to public trust," Edward Lucas, the senior vice president of the Centre for European Policy Analysis told MPs. "And we're not remotely ready for this." Lucas was talking about so-called deepfakes, which he described as "audio and video that look and sound like the real person, saying something that that person never has". Less than three months ago, producing such videos was a laborious process requiring a video editor, vast amounts of reference footage and years of experience. But in the first few months of this year, the technology has exploded into public availability.

Even dogs have dreams, but not you, you are just a machine. Can a robot write a symphony? Can a robot turn a canvas into a beautiful masterpiece?" This famous quote from the film "I, Robot", inspired by Isaac Asimov, the science fiction writer's collection of short stories, poses questions that current technology can already answer. Computer programs that have machine learning capabilities can compose sonatas, songs, and classical pieces, and can even draw pictures at a level on a par with high art.

As we end 2017, I'm tired of writing "lecturing" blogs about what organizations should be doing to master data monetization in order to power their business models and achieve digital transformation. While the objective of every organization should be to master big data and data science (artificial intelligence, machine learning, deep learning) to drive "data monetization," let's take a breath and have some fun. My recent ankle surgery afforded me the opportunity to binge watch "Game of Thrones." As I watched the impending battle between the White Walkers and humanity, I couldn't help but identify a number of lessons that we can learn from Jon Snow's battle with the leader of the White Walkers…and the power of Valyrian steel! Game of Thrones and data, not exactly two things you think are in harmony, but this is where I find myself.

We apply basic statistical reasoning to signal reconstruction by machine learning -- learning to map corrupted observations to clean signals -- with a simple and powerful conclusion: under certain common circumstances, it is possible to learn to restore signals without ever observing clean ones, at performance close or equal to training using clean exemplars. We show applications in photographic noise removal, denoising of synthetic Monte Carlo images, and reconstruction of MRI scans from undersampled inputs, all based on only observing corrupted data.

Metacreation applies tools and techniques from artificial intelligence, artificial life, and machine learning, themselves often inspired by cognitive and natural science, for creative tasks. Musical Metacreation studies the design and use of these generative tools and theories for music making: discovery and exploration of novel musical styles and content, collaboration between human performers and creative software "partners", and design of systems in gaming and entertainment that dynamically generate or modify music. MUME intents to bring together artists, practitioners, and researchers interested in developing systems that autonomously (or interactively) recognize, learn, represent, compose, generate, complete, accompany, or interpret music. It is our pleasure to announce the double special issue on Musical Metacreation recently published the ACM journal Computer in Entertainment. The double issue begins with an introduction to Musical Metacreation that covers the basic for those who are not familiar with the use of generative systems to partially or completely automate musical tasks.

Over the past weeks I have started implementing basic machine learning algorithms in plain Python (Python 3.6). I created the repository to prepare for technical interviews and review my knowledge on algorithms such as k-means, k-nn, logistic regression, neural networks, etc. Also, I wanted to create a knowledge base of easy-to-understand implementations of these algorithms together with the most important theoretical explanations. Some of you might find these implementations helpful when preparing for interviews, starting to learn about machine learning or reviewing basic ML algorithms. I am still working on the repository, so more algorithms will follow over the next months.

The keynote for last years Nvidia GTC conference made headlines around the world for its video featuring of how omnipresent artificial intelligence might be in the future. Playing behind the video was a soundtrack also composed by an AI, AIVA, a small company from Luxembourg. The team behind AIVA are from a classical music background and trained it on music from composers such as Mozart, Beethoven, and Bach to give it an understanding of musical theory, and the patterns behind it. AIVA is purely a composer, providing the music to play, but not actively performing it. So far, this has meant that international orchestras have played its pieces as well as other computer software, such as music for a computer game.

I've been spending some time looking at the convergence behavior of different neural networks trained on MNIST data and cross-entropy loss. I started by training deeper and deeper networks using sigmoid type activations until the learning efficiency got too low before switching to ReLU activations. After switching to ReLU activations, my network converged without too many problems but I noticed that the learning rates exhibited an interesting pattern. In particular, it takes a complete epoch before the loss begins to fall. My weights and biases are initialized uniformly with weights initialized between -0.1 and 0.1.