There has been a lot of research to predict earthquakes and how to increase safety during an earthquake. A question that has remained relatively unexplored is what happens after an earthquake? And how can Artificial Intelligence help? The last question is a big issue to tackle, which we focused on in this challenge. Scientists predict that there will be an earthquake in Istanbul in the near future but the exact date is difficult to identify since Istanbul resides on a fault line.

Interest in edge computing continues to build, as does confusion surrounding the architecture. The situation is similar when it comes to artificial intelligence. The prospect of moving AI to the edge might sound like a recipe for even more confusion. Performing artificial intelligence at the edge is often "just theory quoted in articles," said Martin Davis, managing partner at DUNELM Associates. Still, the concept of edge AI is increasingly hard for industrial and enterprise organizations to ignore.

Welcome to Machine Learning Up-To-Date (ML UTD) 1! The LifeWithData blog separates the signal from the noise in today's hectic front lines of software engineering and machine learning. LifeWithData aims to consistently deliver curated machine learning newsletters that point the reader to key developments without massive amounts of backstory for each. This enables frequent, concise updates across the industry without overloading readers with information. ML UTD 1 brings innovations in the fields of edge computing, deep learning, ML standardization. Let's continue moving away from bulky cloud server costs with Pytorch mobile and SwiftUI.

The future of artificial intelligence (AI) is here: self-driving cars, grocery-delivering drones and voice assistants like Alexa that control more and more of our lives, from the locks on our front doors to the temperatures of our homes. For example, should an autonomous vehicle swerve into a pedestrian or stay its course when facing a collision? These questions plague technology companies as they develop AI at a clip outpacing government regulation, and have led Seattle University to develop a new ethics course for the public. Launched last week, the free, online course for businesses is the first step in a Microsoft-funded initiative to merge ethics and technology education at the Jesuit university. Seattle U senior business-school instructor Nathan Colaner hopes the new course will become a well-known resource for businesses "as they realize that [AI] is changing things," he said.

China's best known AI companies are Sensetime, Megvii, Cloudwalk, Yitu, ByteDance, and the BAT companies -- China's first generation of internet giant: Baidu, Alibaba and Tencent. But there's another giant of artificial intelligence that is rarely discussed in the same breath as the companies mentioned above. The state-owned electric utility monopoly State Grid Corporation of China (hereafter State Grid) is the largest utility company in the world, ranking second on the 2018 Fortune Global 500 List. Less celebrated is that State Grid was the only Chinese company ranked in the top 20 in artificial intelligence (AI) patent applicants, per the World Intellectual Property Organization. In an article (in Chinese) published last year titled "State Grid Corporation of China: A hidden giant in AI," Lǐ Shāng 李熵 gives a portrait of a company whose AI initiatives could change the world.

Articles about how autonomous vehicles are almost here and how big a deal it'll be aren't exactly uncommon, but this one from a couple of days ago over at Interesting Engineering struck me in particular because it rehashes an idea that's at the root of so many articles and self-righteous online arguments from Tesla owners about autonomous cars: that autonomous vehicles will unquestionably be safer than human drivers. I think there's a bit of a logical fallacy there, and there's no reason we have to keep perpetuating this idea. The Interesting Engineering article is called Will Our Children Ever Learn To Drive and is, in turn, inspired by a Motor Trend story about a robotics expert who predicts that kids born today won't ever drive a car, as our "autonomous future is only 10, 15 years out." Now, there's plenty to unpack and discuss right there, as I personally think those timelines are wildly optimistic and don't factor in any number of realities of driving and the world that could affect whether or not your kid chooses to learn to drive, but I want to focus more on one particular part of the IE article. Our children may then have a choice to drive a car, but they will probably never have to drive a car if they don't want to. As a parent myself, I personally hope my children never have to drive a car for one main reason – safety.

Every week, Invector Labs publishes a newsletter that covers the most recent developments in AI research and technology. You can find this week's issue below. You can sign up for it below. Adversarial attacks are a common mechanism to evaluate the robustness of neural networks. Conceptually, adversarial attacks use neural networks to create data samples that disrupt the learning process of other neural networks.

The Gaussian process (GP) model, which has been extensively applied as priors of functions, has demonstrated excellent performance. The specification of a large number of parameters affects the computational efficiency and the feasibility of implementation of a control strategy. We propose a linear model to approximate GPs; this model expands the GP model by a series of basis functions. Several examples and simulation studies are presented to demonstrate the advantages of the proposed method. A control strategy is provided with the proposed linear model. Keywords: Data mining, forecasting, stochastic processes, control strategies INTRODUCTION The Gaussian process (GP) is a powerful modeling tool that has many applications in research and practice. It provides a practical and probabilistic approach to learning in kernel machines. The GP is extensively applied as a prior of a true function.

Semi-supervised learning methods are usually employed in the classification of data sets where only a small subset of the data items is labeled. In these scenarios, label noise is a crucial issue, since the noise may easily spread to a large portion or even the entire data set, leading to major degradation in classification accuracy. Therefore, the development of new techniques to reduce the nasty effects of label noise in semi-supervised learning is a vital issue. Recently, a graph-based semi-supervised learning approach based on Particle competition and cooperation was developed. In this model, particles walk in the graphs constructed from the data sets. Competition takes place among particles representing different class labels, while the cooperation occurs among particles with the same label. This paper presents a new particle competition and cooperation algorithm, specifically designed to increase the robustness to the presence of label noise, improving its label noise tolerance. Different from other methods, the proposed one does not require a separate technique to deal with label noise. It performs classification of unlabeled nodes and reclassification of the nodes affected by label noise in a unique process. Computer simulations show the classification accuracy of the proposed method when applied to some artificial and real-world data sets, in which we introduce increasing amounts of label noise. The classification accuracy is compared to those achieved by previous particle competition and cooperation algorithms and other representative graph-based semi-supervised learning methods using the same scenarios. Results show the effectiveness of the proposed method.

Decision tree learning is a popular classification technique most commonly used in machine learning applications. Recent work has shown that decision trees can be used to represent provably-correct controllers concisely. Compared to representations using lookup tables or binary decision diagrams, decision trees are smaller and more explainable. We present dtControl, an easily extensible tool for representing memoryless controllers as decision trees. We give a comprehensive evaluation of various decision tree learning algorithms applied to 10 case studies arising out of correct-by-construction controller synthesis. These algorithms include two new techniques, one for using arbitrary linear binary classifiers in the decision tree learning, and one novel approach for determinizing controllers during the decision tree construction. In particular the latter turns out to be extremely efficient, yielding decision trees with a single-digit number of decision nodes on 5 of the case studies.