Deep transformer neural network models have improved the predictive accuracy of intelligent text processing systems in the biomedical domain. They have obtained state-of-the-art performance scores on a wide variety of biomedical and clinical Natural Language Processing (NLP) benchmarks. However, the robustness and reliability of these models has been less explored so far. Neural NLP models can be easily fooled by adversarial samples, i.e. minor changes to input that preserve the meaning and understandability of the text but force the NLP system to make erroneous decisions. This raises serious concerns about the security and trust-worthiness of biomedical NLP systems, especially when they are intended to be deployed in real-world use cases. We investigated the robustness of several transformer neural language models, i.e. BioBERT, SciBERT, BioMed-RoBERTa, and Bio-ClinicalBERT, on a wide range of biomedical and clinical text processing tasks. We implemented various adversarial attack methods to test the NLP systems in different attack scenarios. Experimental results showed that the biomedical NLP models are sensitive to adversarial samples; their performance dropped in average by 21 and 18.9 absolute percent on character-level and word-level adversarial noise, respectively. Conducting extensive adversarial training experiments, we fine-tuned the NLP models on a mixture of clean samples and adversarial inputs. Results showed that adversarial training is an effective defense mechanism against adversarial noise; the models robustness improved in average by 11.3 absolute percent. In addition, the models performance on clean data increased in average by 2.4 absolute present, demonstrating that adversarial training can boost generalization abilities of biomedical NLP systems.

When a new computer security vulnerability is publicly disclosed, only a textual description of it is available. Cybersecurity experts later provide an analysis of the severity of the vulnerability using the Common Vulnerability Scoring System (CVSS). Specifically, the different characteristics of the vulnerability are summarized into a vector (consisting of a set of metrics), from which a severity score is computed. However, because of the high number of vulnerabilities disclosed everyday this process requires lot of manpower, and several days may pass before a vulnerability is analyzed. We propose to leverage recent advances in the field of Natural Language Processing (NLP) to determine the CVSS vector and the associated severity score of a vulnerability from its textual description in an explainable manner. To this purpose, we trained multiple BERT classifiers, one for each metric composing the CVSS vector. Experimental results show that our trained classifiers are able to determine the value of the metrics of the CVSS vector with high accuracy. The severity score computed from the predicted CVSS vector is also very close to the real severity score attributed by a human expert. For explainability purpose, gradient-based input saliency method was used to determine the most relevant input words for a given prediction made by our classifiers. Often, the top relevant words include terms in agreement with the rationales of a human cybersecurity expert, making the explanation comprehensible for end-users.

Artificial Intelligence (AI)-based cyber-security methodologies have been developed in response to this unprecedented challenge to help AI information security teams tackle data breaches and cyber security threats. According to a Capgemini Research Institute poll, almost 69% of businesses feel AI plays an active role in combating various types of cybercrimes. Changes in cyber-attacks are rising day by day as technology advances in both directions, and it continues to expand at a rapid pace. Hundreds of billions of time-varying signals must be evaluated depending on the size of your organization in order to correctly quantify risk. As a result, businesses are considering implementing artificial intelligence in cybersecurity. The cyber-world is vast and deep.

U.S. military commanders covered up an air strike over Syria that killed several dozen civilians, dishonestly portraying it as a successful attack against ISIS fighters and ignoring firm recommendations--filed by military lawyers--to investigate the strike as a war crime. The attack and subsequent cover-up--revealed in a long, extensively documented story in this weekend's New York Times--took place in 2019, during the final phase of the U.S. and allied campaign to oust the Islamic State from its self-declared caliphate in Syria. The Times report comes a few months after the final U.S. drone strike in Afghanistan in August, which Pentagon officials touted as halting a terrorist attack--but which in fact, as another Times investigation soon revealed, killed 10 civilians, none of whom had any connection to terrorists. Together, the two reports raise questions about the moral and strategic wisdom of launching airstrikes in areas where civilians and fighters routinely mix. These questions have been raised many times in the course of America's 20-year "global war on terror."

VP Data & AI at ECS, roles have included co-founder at a data analytics startup, VP AI at Booz Allen, and Global Analytics Lead at Accenture. As the influence of artificial intelligence grows, it is increasingly vital to design processes and systems to harness AI while counterbalancing risk. Our charge is to eliminate bias, codify objectives and represent values. Responsible AI ensures alignment to our standards spanning data, algorithms, operations, technology and Human Computer Interaction. I am examining the importance of each of these elements in a series of articles.

Biometrics adoption is being encouraged in the public sector for digital ID and online government applications, as it continues to rise in the private sector from smartphones, where Fingerprint Cards has announced new wins to airport processes, where NEC technology is being deployed and Vision-Box is positioning for more growth. National digital ID programs are under the microscope, while Thales has signed a major deal in Vietnam, and a debate has broken out on facial recognition ethics between Oosto and Clearview AI. The potential for digital identity to boost national economies is examined by the World Economic Forum in a new white paper. The WEF sees digital ID as benefitting people by easing access to a range of services, helping small and medium-sized businesses with easier access to financing, and help establish robust growth in digital service industries, with China's digital wealth-management market offered as an example. Research ICT Africa has released a series of extensive reports delving into the digital identity systems in 10 African countries.

A new machine-learning system helps robots understand and perform certain social interactions. Robots can deliver food on a college campus and hit a hole-in-one on the golf course, but even the most sophisticated robot can't perform basic social interactions that are critical to everyday human life. MIT researchers have now incorporated certain social interactions into a framework for robotics, enabling machines to understand what it means to help or hinder one another, and to learn to perform these social behaviors on their own. In a simulated environment, a robot watches its companion, guesses what task it wants to accomplish, and then helps or hinders this other robot based on its own goals. The researchers also showed that their model creates realistic and predictable social interactions.

There's plenty new in iPadOS 15, but it also features an under-sung accessibility upgrade: support for third-party eye-tracking devices. That'll allow people with disabilities to use iPad apps and speech generation software simply through eye movements -- no touchscreen interaction required. Tobii Dynavox, the assistive tech division of the eye-tracking company Tobii, worked with Apple for years to help make that happen. And now, the firm is ready to announce TD Pilot, a device that aims to bring the iPad experience to the estimated 50 million people globally who need communication assistance. The TD Pilot is basically a super-powered frame for Apple's tablets: It can fit in something as big as the iPad Pro 12.9-inch, and it also packs in large speakers, an extended battery and a wheelchair mount.

Earlier this year, the Indian government issued new guidelines allowing private entities to easily use, create, and access land data instead of going through long clearance protocols. The newly available data includes location information about physical structures, boundaries, natural phenomena, weather patterns, and more, gathered through ground-based survey techniques, photogrammetry using drones, lidar, radar, and so on. On paper, this means a green light for small- and medium-sized companies to collect and use this data to build commercial applications and services related to mapping. It is also a relief to alternative or participatory mapping communities, such as counter-mapping initiatives (in which local, indigenous populations make their own maps in their own contexts), which have so far lurked in a gray area of legality. For the development and academic sectors, too, it heralds greater access to maps and related data for research.

Italian and European authorities weren’t aware that a Chinese state-controlled company bought an Italian military-drone manufacturer and transferred its know-how and technology, revealing how Beijing is skirting weak investment-screening in Europe to acquire sensitive technology.