The four bronze lions that surround Nelson's Column in London's Trafalgar Square are pretty passive-looking creatures. They sit on all fours and gaze blankly ahead, more sphinx than lion. Legend has it that their sculptor originally planned for the animals to be posed in more active stances, stood up on their hind legs and roaring at the square. But Queen Victoria reportedly vetoed the decision as too shocking. Now, 151 years after they were originally unveiled, the lions have a new colleague, and he is definitely turning heads.

There are many ways machines can learn, but for humans nothing beats getting together with like-minded souls who've trodden a similar path. So if you're looking to sharpen up your work in machine learning, artificial intelligence or data science, you should be joining us on Monday morning when we open the doors to Minds Mastering Machines. We've assembled a lineup of speakers who can take you from the fundamentals through to the practical application of key technologies and techniques including TensorFlow, Keras, Lime, GPUs, Deep Learning and Reinforcement Learning. And they'll be discussing how they've applied them in areas such as financial trading, military vehicles, architectural engineering, and just keeping the trains running. We also have some spaces left in our brace of workshops covering developing and deploying machine learning and using the cloud, containers and DevOps to get your project into production.

Salary is £80,000 – £150,000, however, this is up to £200,000 for absolute superstars… we are talking about multiple 1st author publications at NIPS, TPAMI, CVPR, ICML et al. and/or multiple SOTA results across relevant data sets. We are looking for several world-class Senior R&D Scientist's to apply deep learning on a range of bleeding-edge projects that will make AI accessible to the world. This global company's platform has been labelled the most accurate on the planet which has secured them a spot on the big table as a market leader in not only the design but the development of next-gen AI and Computer Vision technology. Working within this environment would allow you to join forces with a team that has 20 years' experience and is recognised as world leading talent when it comes to deep learning & computer vision research The research team is central to making this organisation the best AI company on the planet and work on the core underpinning AI, they continuously aim to push the bar and work with bleeding edge technology & techniques to achieve the impossible. This team look beyond the latest techniques in Deep Learning to what is next.

Chintala's take is that some people would have to be assigned on something like this anyway. If PyTorch had not been created, the other option would be to tweak some existing framework, which would end up requiring the same resources too.

In this, the first episode of the Deep Learning Indaba series, we're joined by Sara Hooker, AI Resident at Google Brain. I had the pleasure of speaking with Sara in the run-up to the Indaba about her work on interpretability in deep neural networks. We discuss what interpretability means and when it's important, and explore some nuances like the distinction between interpreting model decisions vs model function. We also dig into her paper Evaluating Feature Importance Estimates and look at the relationship between this work and interpretability approaches like LIME. We also talk a bit about Google, in particular, the relationship between Brain and the rest of the Google AI landscape and the significance of the recently announced Google AI Lab in Accra, Ghana, being led by friend of the show Moustapha Cisse.

Ruchir Puri is an IBM Fellow and the Chief Architect of IBM Watson. Dr. Puri led Deep Learning and Machine Learning Platform Initiative at IBM Research and also led IBM's efforts in software-hardware acceleration for cognitive and analytic workloads and drove strategy for differentiated cognitive computing infrastructure. He is a Fellow of the IEEE, an ACM Distinguished Speaker, an IEEE Distinguished Lecturer, and was awarded 2014 Asian American Engineer of the Year. Ruchir has been a visiting faculty at Dept. of Computer Science, Stanford Univ...

Yangqing Jia is director of engineering for Facebook's AI platform team, which develops general-purpose open source AI solutions that serve as the backbone of Facebook AI products, such as ranking, computer vision, natural language processing, speech recognition, mobile AI, and AR. He has influential in developing an open source deep learning software stack, many of which serve as the de facto industry standard in AI. Lately, he has been focused on the design ...

Jonathan is passionate about the application of technology towards our environment, human productivity, safety, education, health & longevity. At Intel, Jonathan is responsible for a global team chartered with driving and accelerating innovation and growth in various market segments. His team is a pioneer in artificial intelligence and deep learning applications with computer vision capabilities, supported by software tools & a robust ecosystem. Additionally, Jonathan leads the IoT channel routes-to-market and is responsible for Intel's Chi...

This paper presents a novel physics-informed regularization method for training of deep neural networks (DNNs). In particular, we focus on the DNN representation for the response of a physical or biological system, for which a set of governing laws are known. These laws often appear in the form of differential equations, derived from first principles, empirically-validated laws, and/or domain expertise. We propose a DNN training approach that utilizes these known differential equations in addition to the measurement data, by introducing a penalty term to the training loss function to penalize divergence form the governing laws. Through three numerical examples, we will show that the proposed regularization produces surrogates that are physically interpretable with smaller generalization errors, when compared to other common regularization methods. Introduction Many science and engineering problems require repetitive simulation runs of a model with different input values. Examples of these problems include design optimization, model calibration, sensitivity analysis, what-if analysis, and design space exploration problems. However, in many real-world problems, obtaining a reliable outcome requires large number of these solves (typically for a partial differential equation), which can be prohibitive given the available resources. One way to alleviate this burden is to construct surrogate models [1] that mimic the solution or response surface. One example is building an analytical polynomial function for the displacement of a 2D plate at different locations.

Anomaly detection is often considered a challenging field of machine learning due to the difficulty of obtaining anomalous samples for training and the need to obtain a sufficient amount of training data. In recent years, autoencoders have been shown to be effective anomaly detectors that train only on "normal" data. Generative adversarial networks (GANs) have been used to generate additional training samples for classifiers, thus making them more accurate and robust. However, in anomaly detection GANs are only used to reconstruct existing samples rather than to generate additional ones. This stems both from the small amount and lack of diversity of anomalous data in most domains. In this study we propose MDGAN, a novel GAN architecture for improving anomaly detection through the generation of additional samples. Our approach uses two discriminators: a dense network for determining whether the generated samples are of sufficient quality (i.e., valid) and an autoencoder that serves as an anomaly detector. MDGAN enables us to reconcile two conflicting goals: 1) generate high-quality samples that can fool the first discriminator, and 2) generate samples that can eventually be effectively reconstructed by the second discriminator, thus improving its performance. Empirical evaluation on a diverse set of datasets demonstrates the merits of our approach.