"We're trying to change how people run agronomic research. Instead of establishing a small field plot, running statistics and publishing the means, what we're trying to do involves the farmer far more directly. We are running experiments with farmers' machinery in their own fields. We can detect site-specific responses to different inputs. And we can see whether there's a response in different parts of the field," said Nicolas Martin, assistant professor in the U of I Department of Crop Sciences and co-author of the study.

Federated Learning enables visual models to be trained on-device, bringing advantages for user privacy (data need never leave the device), but challenges in terms of data diversity and quality. Whilst typical models in the datacenter are trained using data that are independent and identically distributed (IID), data at source are typically far from IID. Furthermore, differing quantities of data are typically available at each device (imbalance). In this work, we characterize the effect these real-world data distributions have on distributed learning, using as a benchmark the standard Federated Averaging (FedAvg) algorithm. To do so, we introduce two new large-scale datasets for species and landmark classification, with realistic per-user data splits that simulate real-world edge learning scenarios. We also develop two new algorithms (FedVC, FedIR) that intelligently resample and reweight over the client pool, bringing large improvements in accuracy and stability in training.

Multi-way data analysis has become an essential tool for capturing underlying structures in higher-order datasets stored in tensor $\mathcal{X} \in \mathbb{R} ^{I_1 \times \dots \times I_N} $. $CANDECOMP/PARAFAC$ (CP) decomposition has been extensively studied and applied to approximate $\mathcal{X}$ by $N$ loading matrices $A^{(1)}, \dots, A^{(N)}$ where $N$ represents the order of the tensor. We propose a new efficient CP decomposition solver named NeCPD for non-convex problem in multi-way online data based on stochastic gradient descent (SGD) algorithm. SGD is very useful in online setting since it allows us to update $\mathcal{X}^{(t+1)}$ in one single step. In terms of global convergence, it is well known that SGD stuck in many saddle points when it deals with non-convex problems. We study the Hessian matrix to identify theses saddle points, and then try to escape them using the perturbation approach which adds little noise to the gradient update step. We further apply Nesterov's Accelerated Gradient (NAG) method in SGD algorithm to optimally accelerate the convergence rate and compensate Hessian computational delay time per epoch. Experimental evaluation in the field of structural health monitoring using laboratory-based and real-life structural datasets show that our method provides more accurate results compared with existing online tensor analysis methods.

While automation and digital technologies are disrupting the workplace as we traditionally know it, it has become imperative for organisations to start reskilling their workforce. Automation and artificial intelligence are not only disrupting the assembly lines but right across the so-called blue-collar jobs. This is mainly because in most instances, artificial intelligence (AI), is actually doing a better job than humans. For example, the use of virtual assistants in the workplace is growing. By 2021, Gartner predicts that 25 percent of digital workers will use a virtual employee assistant on a daily basis.

As of March 10, 2020, 467 confirmed cases of COVID-19 have been reported to the Centers for Disease Control and Prevention in the United States. While governments across the globe are working in collaboration with local authorities and health-care providers to track, respond to and prevent the spread of disease caused by the coronavirus, health experts are turning to advanced analytics and artificial intelligence to augment current efforts to prevent further infection. Data and analytics have proved to be useful in combating the spread of disease, and the federal government has access to ample data on the U.S. population's health and travel as well as the migration of both domestic and wild animals -- all of which can be useful in tracking and predicting disease trajectory. Machine learning's ability to consider large amounts of data and offer insights can lead to deeper knowledge about diseases and enable U.S. health and government officials to make better decisions throughout the entire evolution of an outbreak. As the global human population grows and continues to interact with animals, other opportunities for viruses that originate in animals (like COVID-19) could make the jump from to humans and spread.

Following on from my comment in the February issue of DW, I thought it worthwhile to spend a few moments contemplating just how much, or how little, companies (as well as individuals) can learn from the history books. Thankfully, major health scares as per the current coronavirus pandemic, are few and far between. And organisations might do well to spend a little time, after we have weathered the storm, thinking about the future, in terms of the human aspect of their business. Yes, AI and robots are, we hope, immune to illness, but humans are still a vital part of any organisation, and large scale illness is something that might just need a bit more planning for in the future. At least as part of a business continuity/disaster recovery plan, where the emphasis tends to be on the machines, not the humans.

A new industrial revolution is taking place now and AI (AI) is transforming countries economically. The answer to the question of who is ahead and who is behind is determined by the new economic model based on this AI. Dozens of countries, from China to the U.S., from Finland to Kenya, are making significant investments in the area. It should be noted that by 2030, AI studies will generate a gross domestic product (GDP) greater than the current size of the Chinese economy ($15 trillion). From this new economy, China will generate nearly $7 trillion, the U.S. $3.7 trillion, Northern Europe $1.8 trillion, Africa-Oceania $1.2 trillion, the rest of Asia $0.9 trillion and Latin America $0.5 trillion.

With the rapid advances in computing and information technologies, traditional access control models have become inadequate in terms of capturing fine-grained, and expressive security requirements of newly emerging applications. An attribute-based access control (ABAC) model provides a more flexible approach for addressing the authorization needs of complex and dynamic systems. While organizations are interested in employing newer authorization models, migrating to such models pose as a significant challenge. Many large-scale businesses need to grant authorization to their user populations that are potentially distributed across disparate and heterogeneous computing environments. Each of these computing environments may have its own access control model. The manual development of a single policy framework for an entire organization is tedious, costly, and error-prone. In this paper, we present a methodology for automatically learning ABAC policy rules from access logs of a system to simplify the policy development process. The proposed approach employs an unsupervised learning-based algorithm for detecting patterns in access logs and extracting ABAC authorization rules from these patterns. In addition, we present two policy improvement algorithms, including rule pruning and policy refinement algorithms to generate a higher quality mined policy. Finally, we implement a prototype of the proposed approach to demonstrate its feasibility.

Recently, generative adversarial networks (GANs), which learn data distributions through adversarial training, have gained special attention owing to their high image reproduction ability. However, one limitation of standard GANs is that they recreate training images faithfully despite image degradation characteristics such as blur, noise, and compression. To remedy this, we address the problem of blur, noise, and compression robust image generation. Our objective is to learn a non-degraded image generator directly from degraded images without prior knowledge of image degradation. The recently proposed noise robust GAN (NR-GAN) already provides a solution to the problem of noise degradation. Therefore, we first focus on blur and compression degradations. We propose blur robust GAN (BR-GAN) and compression robust GAN (CR-GAN), which learn a kernel generator and quality factor generator, respectively, with non-degraded image generators. Owing to the irreversible blur and compression characteristics, adjusting their strengths is non-trivial. Therefore, we incorporate switching architectures that can adapt the strengths in a data-driven manner. Based on BR-GAN, NR-GAN, and CR-GAN, we further propose blur, noise, and compression robust GAN (BNCR-GAN), which unifies these three models into a single model with additionally introduced adaptive consistency losses that suppress the uncertainty caused by the combination. We provide benchmark scores through large-scale comparative studies on CIFAR-10 and a generality analysis on FFHQ dataset.

Centralized state camera surveillance is but one component of a burgeoning practice of personal data collection paired with artificial intelligence (AI). Camera surveillance is not inherently unlawful and has long been used at border-crossings, airports, and other high-security areas. However, recent technological advances have contributed to the spread of a more intrusive form of video surveillance that includes powerful, if imperfect, facial recognition abilities and AI decision making. While the technology offers states the ability to, among other things, identify lost children, identify criminals, and monitor threats, the new capacity also raises significant human rights issues. The use of camera surveillance has grown with leaps in technology, including the introduction of videocassette recorders in the 1970s and the internet in the 1990s.