Reinforcement learning has traditionally occupied a niche status in the world of artificial intelligence. But reinforcement learning has started to assume a larger role in many AI initiatives in the past few years. Its application sweet spot is in calculation of optimal actions to be taken by agents in environmentally contextualized decision scenarios. Using trial-and-error approaches to maximize an algorithmic reward function, reinforcement learning is well suited to many adaptive-control and multiagent automation applications in IT operations management, energy, health care, commerce, finance, transportation, and finance. And it's being used to train the AI that powers both its traditional focus areas--robotics, gaming, and simulation--and a new generation of AI solutions in edge analytics, natural language processing, machine translation, computer vision, and digital assistants.

The topic of Insurtech is raising growing interest. This is mainly due to the immense size and importance of the insurance market, however, can also be attributed to the promising new opportunities offered by new technologies. As we pointed out in our last column "The Five Insurtech Battles", the applications are very diverse and players in the Insurtech space can be roughly divided into five categories. A unifying trait, however, is that many of these Insurtechs have the common approach to tackling their problems by leveraging data and Artificial Intelligence (AI), which we will discuss in more detail below. Insurance companies have always been very professional and efficient IT organizations compared to other industries and data has always played a major role.

Realising the dream of AI is more process than magic. But the engineering process itself is a journey. Given the talk of AI in the general media, you could forgive yourself for thinking that it will'just happen'. It sometimes sounds as if machine learning is on an unstoppable roll, poised to revolutionise industry and commerce through sheer momentum of technology. But of course, it's not that simple for organisations that want to adopt AI methods.

By developing state-of-the-art algorithms, our team is building models and applications in areas including natural language processing, computer vision, autonomous driving, speech recognition, personalization, anomaly detection, and robotic learning. Many of the standard and benchmark models are developed and provided using our open source neon or Intel nGraph (formerly known as Intel Nervana Graph) AI software platforms and are available on the Model Zoo or as examples in neon. These frameworks take deep learning models through compilation and optimization to run efficiently on a variety of hardware platforms. The models we provide demonstrate the usage and flexibility of Intel frameworks, and enable deep learning researchers and developers to leverage the most relevant topologies and Intel technologies for their use cases. As we move forward with more original research, we will continue to publish papers and open source models.

These days the terms "AI", "Machine Learning", "Deep Learning" are thrown around by companies in every industry, they're the type of words that make any forward-looking executive salivate. You might think these are new concepts that seemed to have appeared overnight, but the reality is they've been around for a while and it's the hard work of many within the field that has really moved it into the spotlight as the latest tech trend. While these terms are sometimes used interchangeably by the media they certainly are not the same, but I'll leave that discussion for another time. It's surely an exciting time for the industry, from a slew of open source libraries (TenserFlow, PredictionIO, DeepLearning4J, or see github) coming into popularity and every cloud provider from Amazon, IBM, Microsoft (the list goes on) all offering their own tools to help get started in the AI/ML/DL field. If you've stumbled on this article, you're probably well aware of everything I've mentioned above, so now that we've gotten past the obligatory intro, let's get to what the title actually claims this article is about.

Deep learning, also known as deep structured learning or hierarchical learning, is a type of machine learning focused on learning data representations and feature learning rather than individual or specific tasks. Feature learning, also known as representation learning, can be supervised, semi-supervised or unsupervised. Deep learning architectures include deep neural networks, deep belief networks and recurrent neural networks. Real-world applications using deep learning include computer vision, speech recognition, machine translation, natural language processing, and image recognition. The following recipe introduces how to implement a deep neural network using TensorFlow, which is an open source software library, originally developed at Google, for complex computation by constructing network graphs of mathematical operations and data (Abadi et al. 2016; Cheng et al. 2017).

Many existing Reinforcement Learning (RL) systems already rely on simulations to explore the solution space and solve complex problems. These include systems based on Self-Play for gaming applications. Self-Play is an essential part of the algorithms used by Google DeepMind in AlphaGo and in the more recent AlphaGo Zero reinforcement learning systems. These are the breakthrough approaches that have defeated the world champion at the ancient Chinese game of Go (D. Silver et al., 2017 https://www.nature.com/articles/nature24270 The newer AlphaGo Zero system has achieved a significant step forward compared to the original Alpha Go system.

Deep learning has shown promising results on many machine learning tasks but DL models are often complex networks with large number of neurons and layers, and recently, complex layer structures known as building blocks. Finding the best deep model requires a combination of finding both the right architecture and the correct set of parameters appropriate for that architecture. In addition, this complexity (in terms of layer types, number of neurons, and number of layers) also present problems with generalization since larger networks are easier to overfit to the data. In this paper, we propose a search framework for finding effective architectural building blocks for convolutional neural networks (CNN). Our approach is much faster at finding models that are close to state-of-the-art in performance. In addition, the models discovered by our approach are also smaller than models discovered by similar techniques. We achieve these twin advantages by designing our search space in such a way that it searches over a reduced set of state-of-the-art building blocks for CNNs including residual block, inception block, inception-residual block, ResNeXt block and many others. We apply this technique to generate models for multiple image datasets and show that these models achieve performance comparable to state-of-the-art (and even surpassing the state-of-the-art in one case). We also show that learned models are transferable between datasets.

Out-of-vocabulary word translation is a major problem for the translation of low-resource languages that suffer from a lack of parallel training data. This paper evaluates the contributions of target-language context models towards the translation of OOV words, specifically in those cases where OOV translations are derived from external knowledge sources, such as dictionaries. We develop both neural and non-neural context models and evaluate them within both phrase-based and self-attention based neural machine translation systems. Our results show that neural language models that integrate additional context beyond the current sentence are the most effective in disambiguating possible OOV word translations. We present an efficient second-pass lattice-rescoring method for wide-context neural language models and demonstrate performance improvements over state-of-the-art self-attention based neural MT systems in five out of six low-resource language pairs.

Impulse response estimation has for a long time been at the core of system identification. Up until some five to seven years ago, the generally held belief in the field was indeed that we knew all there was to know about this topic. However, the enlightening work by Pillonetto and De Nicolao [2010] changed this by showing that the estimate can in fact be improved significantly by assuming a Gaussian Process (GP) prior over the impulse response, which acts as a regularizer. This model-driven approach has since then been further refined [Pillonetto et al., 2011, Chen et al., 2012, Pillonetto et al., 2014], where the prior in this case could be interpreted to encode not only smoothness information, but also information about the exponential decay of the impulse response. In this paper we employ deep leaning (DL) to find a suitable regularizer via a method that is driven by data. Deep learning is a fairly new area of research that continues the work on neural networks from the 1990's. To get a brief, but informative, overview of the field of deep learning we recommend the paper by LeCun et al. [2015] and for a more complete snapshot of the field we refer to the monograph by Goodfel-low et al. [2016]. Deep learning has recently revolutionized several fields, including image recognition (e.g.