Machine-learning phishing webpage detectors (ML-PWD) have been shown to suffer from adversarial manipulations of the HTML code of the input webpage. Nevertheless, the attacks recently proposed have demonstrated limited effectiveness due to their lack of optimizing the usage of the adopted manipulations, and they focus solely on specific elements of the HTML code. In this work, we overcome these limitations by first designing a novel set of fine-grained manipulations which allow to modify the HTML code of the input phishing webpage without compromising its maliciousness and visual appearance, i.e., the manipulations are functionality- and rendering-preserving by design. We then select which manipulations should be applied to bypass the target detector by a query-efficient black-box optimization algorithm. Our experiments show that our attacks are able to raze to the ground the performance of current state-of-the-art ML-PWD using just 30 queries, thus overcoming the weaker attacks developed in previous work, and enabling a much fairer robustness evaluation of ML-PWD.

Artificial intelligence (AI) and machine learning will play a role in protecting the United States from malicious cyber actors. NSA's Jason Wang, technical director for the Computer and Analytic Sciences Research Group, forecasted a future in which AI will support the Intelligence Community's (IC) efforts to secure and defend our Nation's networks. "I think the next frontier for us is probably in the cybersecurity space," he said. "There's a lot of opportunity … to bring machines to this very low latency, highly dynamic problem in ways that really are not human-time kinds of responses." Mr. Wang was responding to a question about cybersecurity and AI posed by former Director of National Intelligence James Clapper, one of several hundred audience members attending a virtual web briefing on July 12th that also featured AI experts from the Central Intelligence Agency (CIA) and National Geospatial Intelligence Agency (NGA).

WASHINGTON -- Winners of the Department of Veterans Affairs 2020-2021 Artificial Intelligence Tech Sprint are six tech companies that created programs aimed at preventing Veteran suicide and improving their health care using the latest AI technology. VA's National Artificial Intelligence Institute competition encourages innovators to develop ways to improve services for Veterans. VA also gave $5,000 awards to JumpStartCSR for an app that integrates with physical therapy to prevent and treat injuries; HIVE Lab at George Washington University for a an app that helps Veterans manage conditions such as diabetes by personalizing treatments based on gut microbiome; and Ouva, LLC for a platform that helps clinicians better monitor vital signs and other health care issues for patients in isolation. The intent of the sprint is to match the private sector with Veterans, VA clinicians and other experts who mentor the companies to brainstorm solutions and new ideas over a three-month period. VA will further evaluate the best ideas and products to potentially adopt at pilot sites and then roll out nationwide.

Electric vehicles are becoming increasingly popular, and self-driving cars are also on the way. When will they be mature enough to meet climate challenges and take to the roads en masse? Writes about the impact of new technologies on society: are we aware of the revolution in progress and its consequences? Many countries are seeking to achieve carbon neutrality within the coming decades. In Europe, the Green DealExternal link has laid down a plan to achieve zero emissions by 2050, and Switzerland has set itself the same deadline. This is an ambitious goal that puts the spotlight on the transport sector, which is responsible for around 16% of global CO2 emissions.External link So what will mobility look like in the future?

We propose Object-oriented Neural Programming (OONP), a framework for semantically parsing documents in specific domains. Basically, OONP reads a document and parses it into a predesigned object-oriented data structure (referred to as ontology in this paper) that reflects the domain-specific semantics of the document. An OONP parser models semantic parsing as a decision process: a neural net-based Reader sequentially goes through the document, and during the process it builds and updates an intermediate ontology to summarize its partial understanding of the text it covers. OONP supports a rich family of operations (both symbolic and differentiable) for composing the ontology, and a big variety of forms (both symbolic and differentiable) for representing the state and the document. An OONP parser can be trained with supervision of different forms and strength, including supervised learning (SL) , reinforcement learning (RL) and hybrid of the two. Our experiments on both synthetic and real-world document parsing tasks have shown that OONP can learn to handle fairly complicated ontology with training data of modest sizes.

We now live in a world where artificial intelligence, and assistive technology is more accessible than ever before. In my previous post'the update round up', I highlight some of the new updates to Apple, Windows, Android and iOS; and how each offering will improve access to content on mobile and desktop devices for various user groups. It's actually closer to hand than we think. Artificial intelligence or (AI), is fast becoming the norm in our daily lives. The first thing to identify is that it doesn't just help people who have additional access requirements, all users regardless of whether or not they use assistive hardware or software benefit from using AI.