Posts about machine learning

Building a Music Recommender with Deep Learning

So just as a quick recap – I started with 9,000 audio files, converted them into 9,000 spectrograms, split them up into 185,000 smaller spectrograms and trained a convolutional neural network on these images. I then extracted 185,000 feature vectors for all these images and calculated the average vector for each of the 9,000 original audio files.

At this point I had now extracted 128 features from the music files that identified different characteristics in the music. So in order to create recommendations of songs that shared similar characteristics, all I needed to find were the vectors that were most similar to one another. To do that, I calculated the cosine similarity between all 9,000 vectors.

Movix.ai - Discover your movie in a few clicks

What is Movix
Movix is a movie recommendation service based on artificial intelligence and Deep Learning. Click movies and tags you like and the system will do the rest — in a few clicks, Movix adapts to your preferences and gives you movies worth watching.

How is this different from other recommedation engines?
Unlike other services, we don't require registration or long history of films you have watched before. Your taste, mood and preferences change every day and we understand that — so every recommendation is completely new! Just click a few movies you would like to watch right now and we will find your movie.

What's inside?
We have designed our own custom Neural Network architecture based on LSTM. It’s built on top of Ternsorflow framework from Google and runs on GPU. Unlike other engines, we have no precalculated recommendations — every click is being processed on

Practical Deep Learning For Coders

Practical Deep Learning For Coders is designed to take anyone with at least one year's coding experience to the point they can apply deep learning best practices to create state of the art models in computer vision, natural language, and recommendation systems. The course is fairly general and students should be able to apply the techniques to other areas as well.

NanoNets: Deep Learning with Limited Data

There has been a recent surge in popularity of Deep Learning, achieving state of the art performance in various tasks like Language Translation, playing Strategy Games and Self Driving Cars requiring millions of data points. One common barrier for using deep learning to solve problems is the amount of data needed to train a model. The requirement of large data arises because of the large number of parameters in the model that machines have to learn.

When working on a problem specific to your domain, often the amount of data needed to build models of this size is impossible to find. However models trained on one task capture relations in the data type and can easily be reused for different problems in the same domain. This technique is referred to as Transfer Learning.

Learn TensorFlow and deep learning, without a Ph.D.

To help more developers embrace deep-learning techniques, without the need to earn a Ph.D., I have attempted to flatten the learning curve by building a short crash-course (3 hours total). The course is focused on a few basic network architectures, including dense, convolutional and recurrent networks, and training techniques such as dropout or batch normalization. (This course was initially presented at the Devoxx conference in Antwerp, Belgium, in November 2016.) By watching the recordings of the course and viewing the annotated slides, you can learn how to solve a couple of typical problems with neural networks and also pick up enough vocabulary and concepts to continue your deep learning self-education — for example, by exploring TensorFlow resources. (TensorFlow is Google’s internally developed framework for deep learning, which has been growing in popularity since it was released as open source in 2015.)

A Course in Machine Learning

Machine learning is the study of algorithms that learn from data and experience. It is applied in a vast variety of application areas, from medicine to advertising, from military to pedestrian. Any area in which you need to make sense of data is a potential consumer of machine learning.

CIML is a set of introductory materials that covers most major aspects of modern machine learning (supervised learning, unsupervised learning, large margin methods, probabilistic modeling, learning theory, etc.). It's focus is on broad applications with a rigorous backbone. A subset can be used for an undergraduate course; a graduate course could probably cover the entire material and then some.