Data's emergence as a critical business asset has been a persistent theme in every Tech Trends report, from the foundational capabilities needed to manage its exploding volumes and complexity to the increasingly sophisticated analytics tools techniques available to unearth business insights from data troves. By harnessing analytics to illuminate patterns, insights, and opportunities hidden within ever-growing data stores, companies have been able to develop new approaches to customer engagement; to amplify employee skills and intelligence; to cultivate new products, services, and offerings; and to explore new business models. Today, more and more CIOs are aggressively laying the foundations needed for their organizations to become more insight-driven. Artificial intelligence (AI)--technologies capable of performing tasks normally requiring human intelligence--is becoming an important component of these analytics efforts. Yet AI is only one part of a larger, more compelling set of developments in the realm of cognitive computing. The bigger story is machine intelligence (MI), an umbrella term for a collection of advances representing a new cognitive era. We are talking here about a number of cognitive tools that have evolved rapidly in recent years: machine learning, deep learning, advanced cognitive analytics, robotics process automation, and bots, to name a few.

The success of various applications including robotics, digital content creation, and visualization demand a structured and abstract representation of the 3D world from limited sensor data. Inspired by the nature of human perception of 3D shapes as a collection of simple parts, we explore such an abstract shape representation based on primitives. Given a single depth image of an object, we present 3D-PRNN, a generative recurrent neural network that synthesizes multiple plausible shapes composed of a set of primitives. Our generative model encodes symmetry characteristics of common man-made objects, preserves long-range structural coherence, and describes objects of varying complexity with a compact representation. We also propose a method based on Gaussian Fields to generate a large scale dataset of primitive-based shape representations to train our network. We evaluate our approach on a wide range of examples and show that it outperforms nearest-neighbor based shape retrieval methods and is on-par with voxel-based generative models while using a significantly reduced parameter space.

Recently, a technique called Layer-wise Relevance Propagation (LRP) was shown to deliver insightful explanations in the form of input space relevances for understanding feed-forward neural network classification decisions. In the present work, we extend the usage of LRP to recurrent neural networks. We propose a specific propagation rule applicable to multiplicative connections as they arise in recurrent network architectures such as LSTMs and GRUs. We apply our technique to a word-based bi-directional LSTM model on a five-class sentiment prediction task, and evaluate the resulting LRP relevances both qualitatively and quantitatively, obtaining better results than a gradient-based related method which was used in previous work.

Synthesizing high-quality images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated by existing text-to-image approaches can roughly reflect the meaning of the given descriptions, but they fail to contain necessary details and vivid object parts. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) to generate 256x256 photo-realistic images conditioned on text descriptions. We decompose the hard problem into more manageable sub-problems through a sketch-refinement process. The Stage-I GAN sketches the primitive shape and colors of the object based on the given text description, yielding Stage-I low-resolution images. The Stage-II GAN takes Stage-I results and text descriptions as inputs, and generates high-resolution images with photo-realistic details. It is able to rectify defects in Stage-I results and add compelling details with the refinement process. To improve the diversity of the synthesized images and stabilize the training of the conditional-GAN, we introduce a novel Conditioning Augmentation technique that encourages smoothness in the latent conditioning manifold. Extensive experiments and comparisons with state-of-the-arts on benchmark datasets demonstrate that the proposed method achieves significant improvements on generating photo-realistic images conditioned on text descriptions.

BigDL runs natively on Apache Spark, and because Qubole offers a greatly enhanced and optimized Spark as a service, it makes for a perfect deployment platform. In this Part 1 of a two-part series, you will learn how to get started with distributed Deep Learning library BigDL on Qubole. By the end, you will have BigDL installed on a Spark cluster with a distributed Deep Learning library readily available for you to use in your Deep Learning applications running on Qubole. In Part 2, you will learn how to write a Deep Learning application on Qubole that uses BigDL to identify handwritten digits (0 to 9) using a LeNet-5 (Convolutional Neural Networks) model that you will train and validate using MNIST database. Before we get started, here's some introduction and background on the technologies involved.

The future of AI lies in neuroscience. So says Google DeepMind's founder Demis Hassabis in a review paper published last week in the prestigious journal Neuron. Hassabis is no stranger to both fields. Armed with a PhD in neuroscience, the computer maverick launched London-based DeepMind to recreate intelligence in silicon. In 2014, Google snagged up the company for over $500 million.

Deep neural networks have many, many learnable parameters that are used to make inferences. Often, this poses a problem in two ways: Sometimes, the model does not make very accurate predictions. It also takes a long time to train them. In a previous post, we covered Train your Deep Learning model faster and sharper: Snapshot Ensembling -- M models for the cost of 1. The authors of this paper propose a method to increase training speed by freezing layers.

Deep learning is applicable to a widening range of artificial intelligence problems, such as image classification, speech recognition, text classification, question answering, text-to-speech, and optical character recognition. It is the technology behind photo tagging systems at Facebook and Google, self-driving cars, speech recognition systems on your smartphone, and much more. Here's a simple example: say you have a large collection of images, and that you want tags associated with each image, for example, "dog," "cat," etc. Deep learning can allow you to create a system that understands how to map such tags to images, learning only from examples. This system can then be applied to new images, automating the task of photo tagging. A deep learning model only has to be fed examples of a task to start generating useful results on new data.

The cognitive and automation technologies behind artificial intelligence (AI) are quietly reshaping the world. Machine learning, deep learning, neural networks, biometrics, natural language processing, Big Data, and predictive analytics have unlocked our imagination, and things that were a dream a few decades ago are now a reality. According to the Gartner's 2016 Hype Cycle, machine learning is now at its peak of inflated expectations, emerging as one of the most innovative and diverse application technologies. Everybody in the tech world is talking about it day and night, though few have actually managed to practice it so far, and the following question certainly comes to mind. Machine learning is a burning subject right now, widely discussed in many articles and at nearly all tech events, but interestingly, it is not new.

Deep learning has been very successful in social sciences and specially areas where there is a lot of data. Trading is another field that can be viewed as social science with a lot of data. With the advent of Deep Learning and Big Data technologies for efficient computation, we are finally able to use the same methods in investment management as we would in face recognition or in making chat-bots. In his session at 20th Cloud Expo, Gaurav Chakravorty, co-founder and Head of Strategy Development at qplum, discussed the transformational impact of Artificial Intelligence and Deep Learning in making trading a scientific process. This focus on learning a hierarchical set of concepts is truly making investing a scientific process, a utility.