We study COMP-AMS, a distributed optimization framework based on gradient averaging and adaptive AMSGrad algorithm. Gradient compression with error feedback is applied to reduce the communication cost in the gradient transmission process. Our convergence analysis of COMP-AMS shows that such compressed gradient averaging strategy yields same convergence rate as standard AMSGrad, and also exhibits the linear speedup effect w.r.t. the number of local workers. Compared with recently proposed protocols on distributed adaptive methods, COMP-AMS is simple and convenient. Numerical experiments are conducted to justify the theoretical findings, and demonstrate that the proposed method can achieve same test accuracy as the full-gradient AMSGrad with substantial communication savings. With its simplicity and efficiency, COMP-AMS can serve as a useful distributed training framework for adaptive gradient methods.

Eileen Yu began covering the IT industry when Asynchronous Transfer Mode was still hip and e-commerce was the new buzzword. Currently an independent business technology journalist and content specialist based in Singapore, she has over 20 years of industry experience with various publications including ZDNet, IDG, and Singapore Press Holdings. If applied inappropriately, artificial intelligence (AI) can bring more harm than good. But, it can offer a much-needed helping hand when humans are unable to find comfort from their own kind. AI hasn't always gotten a good rep.

China has developed a remote sensing satellite powered by the latest artificial intelligence technology that helps the People's Liberation Army (PLA) trace the movements of U.S. aircraft carriers. A new study by Chinese space scientists said the technology was put into use last year in June to detect the movements of the USS Harry S. Truman. The satellite, which has not been named in the study, is said to have alerted Beijing with the precise coordinates of the carrier as it headed to a strait transit drill off the coast of Long Island in New York, reported South China Morning Post. According to the study published by the domestic peer-reviewed journal Spacecraft Engineering last month, the drill held on June 17 involved a joint action of seven warships and planes beside the USS Harry S Truman. Before this satellite, the PLA had to go through a large amount of raw satellite data on the ground to get a clue about such drills happening in the U.S. home waters, and the results usually came after the event was over, the report added. But, with the AI-powered satellites, China could now "live stream" military activities or assets of interest on the other side of the planet, the report quoted the study by space scientist Yang Fang and her colleagues with DFH Satellite.

Notorious facial recognition company Clearview AI has agreed to permanently halt sales of its massive biometric database to all private companies and individuals in the United States as part of a legal settlement with the American Civil Liberties Union, per court records. Monday's announcement marks the close of a two-year legal dispute brought by the ACLU and privacy advocate groups in May of 2020 against the company over allegations that it had violated BIPA, the 2008 Illinois Biometric Information Privacy Act. This act requires companies to obtain permission before harvesting a person's biometric information -- fingerprints, gait metrics, iris scans and faceprints for example -- and empowers users to sue the companies who do not. "Fourteen years ago, the ACLU of Illinois led the effort to enact BIPA – a groundbreaking statute to deal with the growing use of sensitive biometric information without any notice and without meaningful consent," Rebecca Glenberg, staff attorney for the ACLU of Illinois, said in a statement. "BIPA was intended to curb exactly the kind of broad-based surveillance that Clearview's app enables. Today's agreement begins to ensure that Clearview complies with the law. This should be a strong signal to other state legislatures to adopt similar statutes."

Previous work has shown promising results in performing entity linking by measuring not only the affinities between mentions and entities but also those amongst mentions. In this paper, we present novel training and inference procedures that fully utilize mention-to-mention affinities by building minimum arborescences (i.e., directed spanning trees) over mentions and entities across documents in order to make linking decisions. We also show that this method gracefully extends to entity discovery, enabling the clustering of mentions that do not have an associated entity in the knowledge base. We evaluate our approach on the Zero-Shot Entity Linking dataset and MedMentions, the largest publicly available biomedical dataset, and show significant improvements in performance for both entity linking and discovery compared to identically parameterized models. We further show significant efficiency improvements with only a small loss in accuracy over previous work, which use more computationally expensive models.

In day-to-day life, a highly demanding task for IT companies is to find the right candidates who fit the companies' culture. This research aims to comprehend, analyze and automatically produce convincing outcomes to find a candidate who perfectly fits right in the company. Data is examined and collected for each employee who works in the IT domain focusing on their performance measure. This is done based on various different categories which bring versatility and a wide view of focus. To this data, learner analysis is done using machine learning algorithms to obtain learner similarity and developer similarity in order to recruit people with identical working patterns. It's been proven that the efficiency and capability of a particular worker go higher when working with a person of a similar personality. Therefore this will serve as a useful tool for recruiters who aim to recruit people with high productivity. This is to say that the model designed will render the best outcome possible with high accuracy and an immaculate recommendation score.

Many of the proposed machine learning (ML) based network intrusion detection systems (NIDSs) achieve near perfect detection performance when evaluated on synthetic benchmark datasets. Though, there is no record of if and how these results generalise to other network scenarios, in particular to real-world networks. In this paper, we investigate the generalisability property of ML-based NIDSs by extensively evaluating seven supervised and unsupervised learning models on four recently published benchmark NIDS datasets. Our investigation indicates that none of the considered models is able to generalise over all studied datasets. Interestingly, our results also indicate that the generalisability has a high degree of asymmetry, i.e., swapping the source and target domains can significantly change the classification performance. Our investigation also indicates that overall, unsupervised learning methods generalise better than supervised learning models in our considered scenarios. Using SHAP values to explain these results indicates that the lack of generalisability is mainly due to the presence of strong correspondence between the values of one or more features and Attack/Benign classes in one dataset-model combination and its absence in other datasets that have different feature distributions.

Norton says it tracked more than $29 million in bitcoin stolen last year and expects this figure to continue to rise in 2022. The company says in Australia, between January and March of this year, Norton thwarted more than 37,098,261 threats, the equivalent of around 403,241 threats per day. That included 471,361 phishing attempts and 59,540 tech support scams. Its latest Consumer Cyber Safety Pulse report found that deepfakes are being utilised by bad actors to scam consumers and spread disinformation. The Norton Labs team spotted deepfakes used to create fake social media profiles, fuel charity scams and other fraudulent ploys, and spread propaganda relating to the ongoing war in Ukraine.

Secretary shall support a program of fundamental research, development, and demonstration of energy efficient computing and data center technologies relevant to advanced computing applications, including high performance computing, artificial intelligence, and scientific machine learning.").

Multimodal named entity recognition and relation extraction (MNER and MRE) is a fundamental and crucial branch in information extraction. However, existing approaches for MNER and MRE usually suffer from error sensitivity when irrelevant object images incorporated in texts. To deal with these issues, we propose a novel Hierarchical Visual Prefix fusion NeTwork (HVPNeT) for visual-enhanced entity and relation extraction, aiming to achieve more effective and robust performance. Specifically, we regard visual representation as pluggable visual prefix to guide the textual representation for error insensitive forecasting decision. We further propose a dynamic gated aggregation strategy to achieve hierarchical multi-scaled visual features as visual prefix for fusion. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our method, and achieve state-of-the-art performance. Code is available in https://github.com/zjunlp/HVPNeT.