There has been an ongoing cycle where stronger defenses against adversarial attacks are subsequently broken by a more advanced defense-aware attack. We present a new approach towards ending this cycle where we "deflect'' adversarial attacks by causing the attacker to produce an input that semantically resembles the attack's target class. To this end, we first propose a stronger defense based on Capsule Networks that combines three detection mechanisms to achieve state-of-the-art detection performance on both standard and defense-aware attacks. We then show that undetected attacks against our defense often perceptually resemble the adversarial target class by performing a human study where participants are asked to label images produced by the attack. These attack images can no longer be called "adversarial'' because our network classifies them the same way as humans do.

Attribution methods that explains the behaviour of machine learning models, e.g. Convolutional Neural Networks (CNNs), have developed into many different forms, motivated by desirable distinct, though related, criteria. Following the diversity of attribution methods, evaluation tools are in need to answer: which method is better for what purpose and why? This paper introduces a new way to decompose the evaluation for attribution methods into two criteria: ordering and proportionality. We argue that existing evaluations follow an ordering criteria roughly corresponding to either the logical concept of necessity or sufficiency. The paper further demonstrates a notion of Proportionality for Necessity and Sufficiency, a quantitative evaluation to compare existing attribution methods, as a refinement to the ordering criteria. Evaluating the performance of existing attribution methods on explaining the CNN for image classification, we conclude that some attribution methods are better in the necessity analysis and the others are better in the sufficiency analysis, but no method is always the winner on both sides.

One of the best performers on the ASX on Monday was the BrainChip Holdings Ltd (ASX: BRN) share price. The artificial intelligence company's shares rocketed 25% higher to 6.9 cents at one stage before closing the day 14.5% higher. Investors were buying the company's shares after it announced the receipt of an EAR99 classification for its Akida Neuromorphic System-on-Chip (NSoC), Akida Software Development Environment (ADE), and related technologies from the U.S. Government. The Export Administration Regulations (EAR) classification of EAR99, which BrainChip has now formally received, removes the barriers for exporting Akida to non-U.S. The EAR99 designation means the company does not require a pre-approval, or a license from the U.S. Department of Commerce, before delivering its solutions globally as part of sales and market expansion activities.

NEW DELHI/MUMBAI, INDIA – When artist Rachita Taneja heads out to protest in New Delhi, she covers her face with a pollution mask, a hoodie or a scarf to reduce the risk of being identified by police facial recognition software. Police in the Indian capital and the northern state of Uttar Pradesh -- both hotbeds of dissent -- have used the technology during protests that have raged since mid-December against a new citizenship law that critics say marginalizes Muslims. Activists are worried about insufficient regulation around the new technology, amid what they say is a crackdown on dissent under Prime Minister Narendra Modi, whose Hindu nationalist agenda has gathered pace since his re-election in May. "I do not know what they are going to do with my data," said Taneja, 28, who created a popular online cartoon about cheap ways for protesters to hide their faces. "We need to protect ourselves, given how this government cracks down."

Unless you've been living under a rock for the past few years, you will be well aware of the trend towards more conscious consumption. In part recently, this is because of the Greta Thunberg effect, the rise Elon Musk's Tesla electric vehicles and more sustainable forms of transportation such as Bird and Uber JUMP. The future of our planet is very much front of mind for Gen-Z, and our cities (especially our large ones) are some of the most polluted places on earth. Just a few days ago, the UK government announced that the petrol and diesel ban will be brought forward 5 years to 2035 rather than 2040. To add to this, the UK government announced that to accelerate the shift to zero emission cars, all company cars will pay no company car tax in 2020–2021.

Arguably, Artificial intelligence or AI debuted at a conference at Dartmouth University in 1956. At the time, there was a lot of optimism. Some people at the conference believed robots and AI machines would be doing the work of humans by the mid-1970s. Of course, that didn't happen -- what happened instead was that funding dried up and a period called "The AI Winter" began. That ostensibly lasted into the 2000s, when IBM's Watson peaked a lot of interest in artificial intelligence again.

The world never changes quite the way you expect. But at The Verge, we've had a front-row seat while technology has permeated every aspect of our lives over the past decade. Some of the resulting moments -- and gadgets -- arguably defined the decade and the world we live in now. But others we ate up with popcorn in hand, marveling at just how incredibly hard they flopped. This is the decade we learned that crowdfunded gadgets can be utter disasters, even if they don't outright steal your hard-earned cash. It's the decade of wearables, tablets, drones and burning batteries, and of ridiculous valuations for companies that were really good at hiding how little they actually had to offer. Here are 84 things that died hard, often hilariously, to bring us where we are today. Everyone was confused by Google's Nexus Q when it debuted in 2012, including The Verge -- which is probably why the bowling ball of a media streamer crashed and burned before it even came to market.

Machine Learning might be a department of computer science pointed at empowering computers to memorize unused behaviors based on experimental data. The objective is to plan the algorithms that allow a computer to show the behavior learned from past encounters, instead human interaction. Now we will examine applications of machine learning in cybersecurity and see how the machine learning algorithms offering assistance to us for battle with cyber-attacks. Machine learning (without human interaction) can collect analyze and prepare data. In cybersecurity, this innovation makes a big difference to analyze past cyber-attacks and create individual defense reactions.

Dave Aitel is the founder and CTO of Immunity. You can follow him @daveaitel. Export control on AI and machine learning algorithms is becoming a more important part of national security strategy as the world moves to a great-power competition landscape and technological changes force accommodation and rapid change to many national interests. However, like security software before it, AI presents unique challenges to how export control has traditionally worked, and these should be considered before being codified into international regulatory frameworks. As an example, on January 6, 2020, The Bureau of Industry and Security (BIS) in the U.S. Department of Commerce released the following rule, which imposed a license requirement on a particular kind of software useful for automatically identifying objects from drone or other imagery: "Geospatial imagery "software" "specially designed" for training a Deep Convolutional Neural Network to automate the analysis of geospatial imagery and point clouds, and having all of the following: Technical Note: A point cloud is a collection of data points defined by a given coordinate system. A point cloud is also known as a digital surface model."

This study demonstrates that it is possible to generate a highly accurate model to predict inpatient and emergency department utilization using data on socioeconomic determinants of care. ABSTRACT Objectives: To determine if it is possible to risk-stratify avoidable utilization without clinical data and with limited patient-level data. Study Design: The aim of this study was to demonstrate the influences of socioeconomic determinants of health (SDH) with regard to avoidable patient-level healthcare utilization. The study investigated the ability of machine learning models to predict risk using only publicly available and purchasable SDH data. A total of 138,115 patients were analyzed from a deidentified database representing 3 health systems in the United States.