Note: The Statsbot team has already published the article about using time series analysis for anomaly detection. Today, we'd like to discuss time series prediction with a long short-term memory model (LSTMs). We asked a data scientist, Neelabh Pant, to tell you about his experience of forecasting exchange rates using recurrent neural networks. As an Indian guy living in the US, I have a constant flow of money from home to me and vice versa. If the USD is stronger in the market, then the Indian rupee (INR) goes down, hence, a person from India buys a dollar for more rupees.

I've read more than my fair share of articles about all-things AI, including ones on machine learning and deep learning, and two things are clear: (1) my brain is about to short-circuit faster than a vintage robot, and (2) more often than not, the media uses these terms interchangeably even though they're not the same thing. First of all: know that machine learning and deep learning are related and that both fall under the AI umbrella. Think of the three as concentric circles: deep learning is a type of machine learning, machine learning is part of AI, and AI encompasses the entire field of study. In a nutshell, machine learning is an automated process that uses data and math (algorithms) to uncover ("learn") new information without human intervention. Because the machine continuously "self-learns" (or "self-trains"), humans don't need to write code for each process along the way (huge time saver!).

DeepMind, the Google-owned artificial intelligence company, announced the formation of a new research group that will investigate ethical questions surrounding the development of AI and its impact on society. DeepMind Ethics and Society, also known as DMES, will start publishing research papers on a variety of topics relating to the development of artificial intelligence and the potential effects of the technology starting in early 2018. Google has staffed the group with eight full-time researchers to start, including six unpaid external fellows who will partner with academic groups at other institutions who are conducting similar research. DMES will eventually grow to around 25 full-time staffers within the next year. Initial partnerships will include the AI Now Institute at NYU and the Leverhulme Centre for the Future of Intelligence.

This is an excellent book. I have hundreds of papers and books on Neural Nets from the time of Rosenblatt's Perceptron on through autoencoders, recurrent NNs, convolutional NNs, RBM's, DNN's, greedy pretraining, Kolmogrov's universal approximation theorem, optimization methods for weight training, and more. I found this book to provide a conceptual overview of the DNNs and the architectures (feed forward, deep belief, unsupervised pre-trained, convolutional, recurrent, long and short term memory, and recursive, networks). The book provides the conceptual connective tissue that are the muscles that the practitioner must bond to the architectural bones to move forward in Deep Learning. The book is a remarkable debrief by two lead developers of the DL4J framework; Josh Patterson and Adam Gibson.

A new member has just joined the family of Data Science Virtual Machines on Azure: The Deep Learning Virtual Machine. Like other DSVMs in the family, the Deep Learning VM is a pre-configured environment with all the tools you need for data science and AI development pre-installed. The Deep Learning VM is designed specifically for GPU-enabled instances, and comes with a complete suite of deep learning frameworks including Tensorflow, PyTorch, MXNet, Caffe2 and CNTK. All Data Science Virtual Machines, including the Deep Learning Virtual Machine, are available as Windows and Ubuntu Linux instances, and are free of any software charges: pay only for the infrastructure charge according to the power and size of the instance you choose. An Azure account is required, but you can get started with $200 in free Azure credits here.

An NTM is a memory augumented neural network (attached to external memory) where the interactions with the external memory (address, read, write) are done using differentiable transformations. Overall, the network is end-to-end differentiable and thus trainable by a gradient based optimizer. The NTM is processing input in sequences, much like an LSTM, but with additional benfits: (1) The external memory allows the network to learn algorithmic tasks easier (2) Having a larger capacity without increasing the network's trainable parameters. The external memory allows the NTM to learn algorithmic tasks, that are much harder for LSTM to learn, and to maintain an internal state much longer than traditional LSTMs. This repository implements a vanilla NTM in a straight forward way.

Researchers at The Alan Turing Institute in the United Kingdom are using artificial intelligence and machine learning to push the state of the art in data science to better understand what makes us happy, angry and frustrated. "Our research seeks to try and measure aspects of the world that we, as humans, are hugely aware of but that traditionally we've had near to no numbers on," says Suzy Moat, an associate professor of behavioral science at Warwick Business School and a Turing Fellow. For example, many people believe that the environment they spend their time in profoundly impacts their day-to-day well-being. Until recently, however, scientists lacked the data to test this hypothesis, Moat explained. Now, the world is awash in data along with the computing resources to quantitatively analyze what settings make people happy.

By its cofounder's reckoning, Efinix is in the right place at the right time. Engineers are struggling to squeeze AI, and especially its deep learning variant, into chips where cost and power are real constraints. The startup, based in Santa Clara, Calif., plans to deliver a new kind of field-programmable gate array (FPGA) technology that is about one-quarter the size of comparable chips, consumes half the power, and is considerably less complex to construct. That combination--what Efinix calls Quantum programmable technology--could help push deep learning and AI in general away from central computers and servers and out toward where the data they work on is being generated, according to cofounder, president, and CEO Sammy Cheung. FPGA's have been using the same basic architecture for decades.

As we delve deeper into the age of digital transformation, machine learning has been a revolutionizing element during this technology shift. The capabilities machine learning is achieving are exciting-- yet, almost unimaginable. Google is a key player in machine learning technology, as the company is making major investments in the space with new APIs and the acquisition of Kaggle, one of the largest and most active communities for data scientists. "Research at Google is at the forefront of innovation in machine intelligence, with active research exploring virtually all aspects of machine learning, including deep learning and more classical algorithms" (Google). Below are some recent advancements in machine learning launched by Google that are sure to reshape the way you work and make things a little bit easier.

To understand why WaveNet improves on the current state of the art, it is useful to understand how text-to-speech (TTS) - or speech synthesis - systems work today. The majority of these are based on so-called concatenative TTS, which uses a large database of high-quality recordings, collected from a single voice actor over many hours. These recordings are split into tiny chunks that can then be combined - or concatenated - to form complete utterances as needed. However, these systems can result in unnatural sounding voices and are also difficult to modify because a whole new database needs to be recorded each time a set of changes, such as new emotions or intonations, are needed. To overcome some of these problems, an alternative model known as parametric TTS is sometimes used.