The spatio-temporal graph learning is becoming an increasingly important object of graph study. Many application domains involve highly dynamic graphs where temporal information is crucial, e.g. traffic networks and financial transaction graphs. Despite the constant progress made on learning structured data, there is still a lack of effective means to extract dynamic complex features from spatio-temporal structures. Particularly, conventional models such as convolutional networks or recurrent neural networks are incapable of revealing the temporal patterns in short or long terms and exploring the spatial properties in local or global scope from spatio-temporal graphs simultaneously. To tackle this problem, we design a novel multi-scale architecture, Spatio-Temporal U-Net (ST-UNet), for graph-structured time series modeling. In this U-shaped network, a paired sampling operation is proposed in spacetime domain accordingly: the pooling (ST-Pool) coarsens the input graph in spatial from its deterministic partition while abstracts multi-resolution temporal dependencies through dilated recurrent skip connections; based on previous settings in the downsampling, the unpooling (ST-Unpool) restores the original structure of spatio-temporal graphs and resumes regular intervals within graph sequences. Experiments on spatio-temporal prediction tasks demonstrate that our model effectively captures comprehensive features in multiple scales and achieves substantial improvements over mainstream methods on several real-world datasets.

Because the choice and tuning of the optimizer affects the speed, and ultimately the performance of deep learning, there is significant past and recent research in this area. Yet, perhaps surprisingly, there is no generally agreed-upon protocol for the quantitative and reproducible evaluation of optimization strategies for deep learning. We suggest routines and benchmarks for stochastic optimization, with special focus on the unique aspects of deep learning, such as stochasticity, tunability and generalization. As the primary contribution, we present DeepOBS, a Python package of deep learning optimization benchmarks. The package addresses key challenges in the quantitative assessment of stochastic optimizers, and automates most steps of benchmarking. The library includes a wide and extensible set of ready-to-use realistic optimization problems, such as training Residual Networks for image classification on ImageNet or character-level language prediction models, as well as popular classics like MNIST and CIFAR-10. The package also provides realistic baseline results for the most popular optimizers on these test problems, ensuring a fair comparison to the competition when benchmarking new optimizers, and without having to run costly experiments. It comes with output back-ends that directly produce LaTeX code for inclusion in academic publications. It supports TensorFlow and is available open source.

How to generate semantically meaningful and structurally sound adversarial examples? We propose to answer this question by restricting the search for adversaries in the true data manifold. To this end, we introduce a stochastic variational inference method to learn the data manifold, in the presence of continuous latent variables with intractable posterior distributions, without requiring an a-priori form for the data underlying distribution. We then propose a manifold perturbation strategy that ensures the cases we perturb remain in the manifold of the original examples and thereby generate the adversaries. We evaluate our approach on a number of image and text datasets. Our results show the effectiveness of our approach in producing coherent, and realistic-looking adversaries that can evade strong defenses known to be resilient to traditional adversarial attacks.

Multi-scale deep CNN architecture [1, 2, 3] successfully captures both fine and coarse level image descriptors for visual similarity task, but they come up with expensive memory overhead and latency. In this paper, we propose a competing novel CNN architecture, called MILDNet, which merits by being vastly compact (about 3 times). Inspired by the fact that successive CNN layers represent the image with increasing levels of abstraction, we compressed our deep ranking model to a single CNN by coupling activations from multiple intermediate layers along with the last layer. Trained on the famous Street2shop dataset [4], we demonstrate that our approach performs as good as the current state-of-the-art models with only one third of the parameters, model size, training time and significant reduction in inference time. The significance of intermediate layers on image retrieval task has also been shown to be performing on popular datasets Holidays, Oxford, Paris [5]. So even though our experiments are done on ecommerce domain, it is applicable to other domains as well. We further did an ablation study to validate our hypothesis by checking the impact on adding each intermediate layer. With this we also present two more useful variants of MILDNet, a mobile model (12 times smaller) for on-edge devices and a compactly featured model (512-d feature embeddings) for systems with less RAMs and to reduce the ranking cost. Further we present an intuitive way to automatically create a tailored in-house triplet training dataset, which is very hard to create manually. This solution too can also be deployed as an all-inclusive visual similarity solution. Finally, we present our entire production level architecture which currently powers visual similarity at Fynd.

Several years ago, the bank I worked at installed an artificial intelligence (AI) system that detected potential credit card fraud. The system worked well and detected and broke up several fraud attempts. However, just as we were feeling pretty comfortable with it, we received a phone call from a very irate board member. The board member attempted to make a large purchase at a home improvement store, and his credit card was denied. The situation was inconvenient and embarrassing. SEE: IT leader's guide to deep learning (Tech Pro Research) What we learned, unfortunately, was that AI systems are capable of issuing false positives, or inaccurate results.

OpenAI, an artificial intelligence research group created by Silicon Valley investors as a non-profit, will now be seeking "capped" profit, according to a blog post on the OpenAI website published Monday. SpaceX and Tesla CEO Elon Musk, startup accelerator Y Combinator president Sam Altman, and several other Silicon Valley figures launched OpenAI in late 2015 with $1 billion in seed funding and the stated goal of ensuring that AI "benefits all of humanity." Musk stepped down from OpenAI in February 2018. Since its founding, the group has conducted research with reinforcement learning, robotics, and language. According to OpenAI, the original nonprofit entity will own a limited partnership called OpenAI LP that's designed to give a "capped return" to investors and employees and funnel excess funds back to the nonprofit.

The field of computer vision is shifting from statistical methods to deep learning neural network methods. There are still many challenging problems to solve in computer vision. Nevertheless, deep learning methods are achieving state-of-the-art results on some specific problems. It is not just the performance of deep learning models on benchmark problems that is most interesting; it is the fact that a single model can learn meaning from images and perform vision tasks, obviating the need for a pipeline of specialized and hand-crafted methods. In this post, you will discover nine interesting computer vision tasks where deep learning methods are achieving some headway.

OpenAI today announced the creation of OpenAI LP, a for-profit company that will be owned and controlled by the OpenAI nonprofit organization's board of directors. The new Delaware-based limited partnership was created to speed progress toward OpenAI's goal of advancing AI and eventually creating safe artificial general intelligence (AGI) system. OpenAI LP plans to raise and invest billions of dollars in the years ahead. Unlike narrow artificial intelligence common today, which can predict the probability of outcomes or recommend content in your Facebook News Feed, OpenAI defines AGI as a highly autonomous system able to outperform humans at most tasks. Sam Altman will serve as CEO of the new entity, while Greg Brockman will act as CTO and Ilya Sutskever as chief scientist.

You can modify the code and tune hyperparameters to get instant feedback to accumulate practical experiences in deep learning.

As with previous Summits, there'll be leading data scientists there, highlighting how they use KNIME Software for solving data problems across industries such as telecommunications, retail, life sciences, manufacturing, finance, and more. We've also got our KNIME courses on offer, giving you the chance to extend your KNIME knowledge, plus an exciting social program, providing plenty of networking opportunities. On March 18 and 19 we are offering several one day KNIME Courses that cover a variety of topics. We'll also be running a KNIME Server half-day workshop on Friday, March 22. Watch this space for more details. We'll take a step back in time and look at the last four versions of KNIME Analytics Platform and all the neat features that have been released.