The academic Deep Learning research community has largely stayed away from the financial markets. Maybe that's because the finance industry has a bad reputation, the problem doesn't seem interesting from a research perspective, or because data is difficult and expensive to obtain. In this post, I'm going to argue that training Reinforcement Learning agents to trade in the financial (and cryptocurrency) markets can be an extremely interesting research problem. I believe that it has not received enough attention from the research community but has the potential to push the state-of-the art of many related fields. It is quite similar to training agents for multiplayer games such as DotA, and many of the same research problems carry over. Knowing virtually nothing about trading, I have spent the past few months working on a project in this field. This is not a "price prediction using Deep Learning" post. So, if you're looking for example code and models you may be disappointed. Instead, I want to talk on a more high level about why learning to trade using Machine Learning is difficult, what some of the challenges are, and where I think Reinforcement Learning fits in. If there's enough interest in this area I may follow up with another post that includes concrete examples. I expect most readers to have no background in trading, just like I didn't, so I will start out with covering some of the basics.

It's a few years old, but still acts as a great technical introduction to the topic.

Google recently added the Tensor Processing Unit v2 (TPUv2), a custom-developed microchip to accelerate deep learning, to its cloud offering. The TPUv2 is the second generation of this chip and the first publicly available deep learning accelerator that has the potential of becoming an alternative to Nvidia GPUs. We recently reported our first experience and received a lot of requests for a more detailed comparison to Nvidia V100 GPUs. Providing a balanced and meaningful comparison for deep learning accelerators is not a trivial task. Due to the future importance of this product category and the lack of detailed comparisons we felt the need to create one on our own.

As we mentioned above, Facebook worked with Microsoft to launch the ONNX standard, but it also has its own Caffe2 deep learning framework. It uses Caffe2 AI technology to power much of the backend on Facebook, including automatic removal of content that violates terms of use and language translations. And then there are all those Facebook chatbots, which businesses can use to interact with users in Messenger.

To unlock insights that deliver significant business performance improvement, applying artificial intelligence and deep learning techniques produce profound outcomes that drive asset management improvements. To learn more about how this translates into value at bottom line, get in touch with us via: enquiries@drivingfueliq.com or visit www.drivingfueliq.com

Using already existing models in ML/DL libraries might be helpful in some cases. But to have better control and understanding, you should try to implement them yourself. This article shows how a CNN is implemented just using NumPy. Convolutional neural network (CNN) is the state-of-art technique for analyzing multidimensional signals such as images. There are different libraries that already implements CNN such as TensorFlow and Keras.

Part of the magic sauce for making the deep learning models work in production is regularization. For this blog post I'll use definition from Ian Goodfellow's book: regularization is "any modification we make to the learning algorithm that is intended to reduce the generalization error, but not its training error". For better theoretical understanding, I'd recommend checking out the chapter of the deep learning book dedicated to regularization. Generalization in machine learning refers to how well the concepts learned by the model apply to examples which were not seen during training. The goal of most machine learning models is to generalize well from the training data, in order to make good predictions in the future for unseen data. Overfitting happens when the models learns too well the details and the noise from training data, but it doesn't generalize well, so the performance is poor for testing data.

Code for this project can be found on: Github. I've also written about this project on Medium and Quora. This project was featured in the Towards Data Science publication here. In the third project of the udacity self-driving car engineer course, I used a car simulator provided by the course organizers to collect data on expected driving behaviour, then used that data to train a deep learning CNN model (NVIDIA convolutional neural network architecture) to drive the car in the simulated environment. The simulator can be run in two modes: training mode and autonomous mode.

Artificial intelligence (AI) is no longer the stuff of science fiction. The majority of businesses today are using AI in some form, and those that aren't have plans to in the near future. Deep learning, a technique that's largely responsible for the widespread adoption of AI, has gained particular momentum as of late, with leading companies like Google, Microsoft and Amazon introducing deep learning across their services and replacing their existing machine learning systems with deep learning-based models. A recent O'Reilly study on how businesses are putting AI to work through deep learning, found that 54 percent of businesses predict deep learning will play a large or essential role in their future projects. Another 38 percent expect to use some amount of deep learning, and only eight percent of businesses said deep learning wouldn't play a role in their future projects.

Sequence prediction is one of the hottest application of Deep Learning these days. From building recommendation systems to speech recognition and natural language processing, its potential is seemingly endless. This is enabling never-thought-before solutions to emerge in the industry and is driving innovation. There are many different ways to perform sequence prediction such as using Markov models, Directed Graphs etc. from the Machine Learning domain and RNNs/LSTMs from the Deep Learning domain. In this article, we will see how we can perform sequence prediction using a relatively unknown algorithm called Compact Prediction Tree (CPT).