This paper presents OptNet, a network architecture that integrates optimization problems (here, specifically in the form of quadratic programs) as individual layers in larger end-to-end trainable deep networks. These layers encode constraints and complex dependencies between the hidden states that traditional convolutional and fully-connected layers often cannot capture. In this paper, we explore the foundations for such an architecture: we show how techniques from sensitivity analysis, bilevel optimization, and implicit differentiation can be used to exactly differentiate through these layers and with respect to layer parameters; we develop a highly efficient solver for these layers that exploits fast GPU-based batch solves within a primal-dual interior point method, and which provides backpropagation gradients with virtually no additional cost on top of the solve; and we highlight the application of these approaches in several problems. In one notable example, we show that the method is capable of learning to play mini-Sudoku (4x4) given just input and output games, with no a priori information about the rules of the game; this highlights the ability of our architecture to learn hard constraints better than other neural architectures.

We consider the task of weakly supervised one-shot detection. In this task, we attempt to perform a detection task over a set of unseen classes, when training only using weak binary labels that indicate the existence of a class instance in a given example. The model is conditioned on a single exemplar of an unseen class and a target example that may or may not contain an instance of the same class as the exemplar. A similarity map is computed by using a Siamese neural network to map the exemplar and regions of the target example to a latent representation space and then computing cosine similarity scores between representations. An attention mechanism weights different regions in the target example, and enables learning of the one-shot detection task using the weaker labels alone. The model can be applied to detection tasks from different domains, including computer vision object detection. We evaluate our attention Siamese networks on a one-shot detection task from the audio domain, where it detects audio keywords in spoken utterances. Our model considerably outperforms a baseline approach and yields a 42.6% average precision for detection across 10 unseen classes. Moreover, architectural developments from computer vision object detection models such as a region proposal network can be incorporated into the model architecture, and results show that performance is expected to improve by doing so.

This paper proposes a fully connected neural network model to map samples from a uniform distribution to samples of any explicitly known probability density function. During the training, the Jensen-Shannon divergence between the distribution of the model's output and the target distribution is minimized. We experimentally demonstrate that our model converges towards the desired state. It provides an alternative to existing sampling methods such as inversion sampling, rejection sampling, Gaussian mixture models and Markov-Chain-Monte-Carlo. Our model has high sampling efficiency and is easily applied to any probability distribution, without the need of further analytical or numerical calculations. It can produce correlated samples, such that the output distribution converges faster towards the target than for independent samples. But it is also able to produce independent samples, if single values are fed into the network and the input values are independent as well. We focus on one-dimensional sampling, but additionally illustrate a two-dimensional example with a target distribution of dependent variables.

Intel recently released its BigDL project for distributed deep learning on Apache Spark. BigDL has native Spark integration, allowing it to leverage Spark during model training, prediction, and tuning. This blog post gives highlights of BigDL and a tutorial showing how to get started with BigDL on Databricks. BigDL is an open source deep learning library from Intel. Modeled after Torch, BigDL provides functionality both for low-level numeric computing and high-level neural networks.

Within three years deep learning will change front-end development. It will increase prototyping speed and lower the barrier for building software. The field took off last year when Tony Beltramelli introduced the pix2code paper and Airbnb launched sketch2code. Currently, the largest barrier to automating front-end development is computing power. However, we can use current deep learning algorithms, along with synthesized training data, to start exploring artificial front-end automation right now.

The work of researchers at the University of Wyoming was featured in a year-end review of content produced by a prestigious international journal. In 2017, the video team for the journal Science created nearly 180 videos on various topics. The most-viewed entry for the entire year accompanied a special package on artificial intelligence and featured Science staff writer Paul Voosen. The basis for the video, "A.I. detectives are cracking open the black box of deep learning," came largely from the work of a team of researchers, two of whom are associated with UW: Department of Computer Science Associate Professor Jeff Clune and graduate student Anh Nguyen, who now is an assistant professor at Auburn University. Science's video explores information from a video, titled "Deep Visualization Toolbox" (www.youtube.com/watch?v AgkfIQ4IGaM) and a paper, "Understanding neural networks through deep visualization," both including contributions from Clune and Nguyen, along with Cornell University's Jason Yosinski and Hod Lipson, and California Institute of Technology's Tom Fuchs.

Artificial Intelligence, Machine Learning and Deep Learning are all the rage in the press these days, and if you want to be a good Data Scientist you're going to need more than just a passing understanding of what they are and what you can do with them. There are loads of different methodologies, but for me I would always suggest Artificial Neural Networks as the first AI to learn - but then I've always had a soft spot for ANNs since I did my PhD on them. They've been around since the 1970s, and until recently have only really been used as research tools in medicine and engineering. Google, Facebook and a few others, though, have realised that there are commercial uses for ANNs, and so everyone is interested in them again. When it comes to algorithms used in AI, Machine Learning and Deep Learning, there are 3 types of learning process (aka'training').

Artificial Intelligence, Machine Learning and Deep Learning (AI ML DL) are at the heart of digital transformation by enabling organizations to exploit their growing wealth of big data to optimize key business and operational use cases. See "Artificial Intelligence is not Fake Intelligence" for more details on AI ML DL. And the business ramifications are staggering! BusinessWeek (October 23, 2017) reported a dramatic increase in mentions of "artificial intelligence" during 363 third quarter earnings calls (see Figure 2). To help our clients exploit the business and operational benefits of AI ML DL, Dell EMC has created "Ready Bundles" that are designed to simplify the configuration, deployment and management of AI ML DL solutions.

OurCrowd, an Israeli equity crowdfunding platform, launched on Thursday a $100 million global fund to finance early-stage companies focused on artificial intelligence (AI), deep-learning, Internet of Things (IoT) and robotics.

The year 2017 was a seminal year for the field of Artificial Intelligence (AI). Suddenly, everyone everywhere was talking about AI, what it means, and how it will affect human societies and economies. Jobs, warfare, healthcare, film-making, even art--no area of human enterprise seemed to be immune from discussions of the coming machine onslaught. Overall, there were three very important outcomes for the field of AI in 2017. First, technologically, the single most important breakthrough in 2017 was the development of Google Deep Mind's AlphaGo Zero. AlphaGo Zero built on the earlier astonishing success of the AlphaGo program, which mastered the game of Go--an East Asian game widely believed to be significantly more complex than chess.