In early March 2020, UK artificial intelligence (AI) security startup Darktrace was able to contain the spread of a sophisticated attack by Chinese cyber espionage and cyber crime group APT41 exploiting a zero-day vulnerability in Zoho ManageEngine. In a blog post describing the attack, Max Heinemeyer, director of threat hunting at Darktrace, wrote: "Without public indicators of compromise (IoCs) or any open source intelligence available, targeted attacks are incredibly difficult to detect. Even the best detections are useless if they cannot be actioned by a security analyst at an early stage. Too often, this occurs because of an overwhelming volume of alerts, or simply because the skills barrier to triage and investigation is too high." Heinemeyer says Darktrace's Cyber AI platform was able to detect the subtle signs of this targeted, unknown attack at an early stage, without relying on prior knowledge.

Toronto and the corridor that stretches west to Kitchener and Waterloo is already Canada's capital of finance and technology--and naturally, the region's leaders want to set an example for the rest of the world. That's part of the reason why in 2017, municipal organizations in Toronto tapped Google's sister company Sidewalk Labs to redevelop a disused waterfront industrial district as a high-tech prototype for the "smarter, greener, more inclusive cities" of tomorrow. But within three years the deal had collapsed, a victim of conflicting visions, public concerns over privacy and surveillance, and (to hear Sidewalk Labs tell it) pandemic-era economic change. Journalist Brian Barth, who trained in urban planning and spent seven years living and working in Toronto before returning to the US this summer, says the Sidewalk fiasco also symbolizes a larger difference: the contrast between Silicon Valley's hard-charging, individualist, libertarian ethos and a Canadian business style that emphasizes collaboration, respect, and social responsibility. In this edition of Deep Tech, Barth talks about the tensions that led to Sidewalk Labs' departure and the strategies Canadian CEOs are following to build a more open and inclusive tech sector. Toronto would like to be seen as the nice person's Silicon Valley, if that's not too much trouble, June 17, 2020 Wade Roush: Is Toronto like Silicon Valley for nice people?

After clearing the institute's security, I was told to wait in a lobby monitored by cameras. On its walls were posters of China's most consequential postwar leaders. He looked serene, as though satisfied with having freed China from the Western yoke. Next to him was a fuzzy black-and-white shot of Deng Xiaoping visiting the institute in his later years, after his economic reforms had set China on a course to reclaim its traditional global role as a great power. The lobby's most prominent poster depicted Xi Jinping in a crisp black suit.

Back on October 12, 2019, the world witnessed a previously unimaginable accomplishment- the first sub-two-hour marathon was run in an incredible time of 1:59:40 by Kenyan native Eliud Kipchoge. He would later say in regards to the amazing achievement that he "expected more people all over the world to run under 2 hours after today" [1]. While Kipchoge set new records in long distance running, across the world a team of natural language processing (NLP) experts at OpenAI, the Elon Musk-backed AI firm, published a new transformer-based language model with 1.5 billion parameters that achieved previously unthinkable performance in nearly every language task it faced [2]. The main takeaway from the paper by many experts was that bigger is better-the intelligence of transformer models can dramatically increase with the scale of parameters. In March of 2020, this theory gained support with OpenAI's release of version three of the model or GPT-3 which encapsulates a staggering 175 billion parameters and achieved even more remarkable performance than version 2, despite sharing, quite literally, the same architecture [3].

Shanghai/Beijing – Huawei Technologies founder Ren Zhengfei's global ambitions are marked in bricks and mortar at a new company campus in southern China, where the buildings are replicas from European cities. Zhang Yiming, founder of ByteDance, the operator of short video app TikTok, has plastered his Beijing headquarters with posters including a cover of former Google CEO Eric Schmidt's book "How Google Works," and has long said he will build a global firm that can compete with U.S. tech giants. But the two companies that best exemplify China's ambitions to challenge U.S. tech dominance are now stymied by strains in relations between China and countries including the United States, India, Australia and Britain. Chinese companies with world-beating technology -- including drone-maker DJI, artificial intelligence firms Megvii, SenseTime and iFlytek, surveillance camera vendor Hikvision and e-commerce conglomerate Alibaba Group -- are also among those losing access to markets. Smaller companies are being forced to rethink too. "What we are experiencing now is unprecedented," said a Chinese startup founder who has operations in the United States and India but asked not to be identified as he is now considering walking away.

Washington – The United States and close ally Australia held high-level talks on China on Tuesday and agreed on the need to uphold a rules-based global order, but the Australian foreign minister stressed that Canberra's relationship with China was important and it had no intention of injuring it. U.S. Secretary of State Mike Pompeo and Defense Secretary Mark Esper held two days of talks in Washington with their Australian counterparts, Foreign Minister Marise Payne and Defense Minister Linda Reynolds, who had flown around the world for the meetings despite the COVID-19 pandemic and face two weeks of quarantine on their return. At a joint news conference, Pompeo praised Australia for standing up to pressure from China and said Washington and Canberra would continue to work together to reassert the rule of law in the South China Sea, where China has been pressing its claims. Payne said the United States and Australia shared a commitment to the rule of law and had reiterated their commitment to hold countries to account for breaches, such as China's erosion of freedoms in Hong Kong. She said the two sides had also agreed to form a working group to monitor and respond to harmful disinformation and would look at ways to expand cooperation on infectious diseases, including access to vaccines.

A new study has found that the masks which protect people from spreading the coronavirus also have a second use, breaking facial recognition algorithms. Researchers from the National Institute of Standards and Technology have found that the best facial recognition algorithms had significantly higher error rates when trying to identify someone wearing a cloth covering. The researchers tested one-to-one matching algorithms, where a photo is compared to a different photo of the same person. This verification method is commonly used to unlock smartphones, or check passports. It drew digital masks onto the faces in a trove of border crossing photographs, and then compared those photos against another database of unmasked people seeking visas and other immigration benefits.

As machine learning is increasingly used to help make decisions, there is a demand for these decisions to be explainable. Arguably, the most explainable machine learning models use decision rules. This paper focuses on decision sets, a type of model with unordered rules, which explains each prediction with a single rule. In order to be easy for humans to understand, these rules must be concise. Earlier work on generating optimal decision sets first minimizes the number of rules, and then minimizes the number of literals, but the resulting rules can often be very large. Here we consider a better measure, namely the total size of the decision set in terms of literals. So we are not driven to a small set of rules which require a large number of literals. We provide the first approach to determine minimum-size decision sets that achieve minimum empirical risk and then investigate sparse alternatives where we trade accuracy for size. By finding optimal solutions we show we can build decision set classifiers that are almost as accurate as the best heuristic methods, but far more concise, and hence more explainable.

Architectures that implement the Common Model of Cognition - Soar, ACT-R, and Sigma - have a prominent place in research on cognitive modeling as well as on designing complex intelligent agents. In this paper, we explore how computational models of analogical processing can be brought into these architectures to enable concept acquisition from examples obtained interactively. We propose a new analogical concept memory for Soar that augments its current system of declarative long-term memories. We frame the problem of concept learning as embedded within the larger context of interactive task learning (ITL) and embodied language processing (ELP). We demonstrate that the analogical learning methods implemented in the proposed memory can quickly learn a diverse types of novel concepts that are useful not only in recognition of a concept in the environment but also in action selection. Our approach has been instantiated in an implemented cognitive system \textsc{Aileen} and evaluated on a simulated robotic domain.

Local Interpretable Model-Agnostic Explanations (LIME) is a popular method to perform interpretability of any kind of Machine Learning (ML) model. It explains one ML prediction at a time, by learning a simple linear model around the prediction. The model is trained on randomly generated data points, sampled from the training dataset distribution and weighted according to the distance from the reference point - the one being explained by LIME. Feature selection is applied to keep only the most important variables. LIME is widespread across different domains, although its instability - a single prediction may obtain different explanations - is one of the major shortcomings. This is due to the randomness in the sampling step, as well as to the flexibility in tuning the weights and determines a lack of reliability in the retrieved explanations, making LIME adoption problematic. In Medicine especially, clinical professionals trust is mandatory to determine the acceptance of an explainable algorithm, considering the importance of the decisions at stake and the related legal issues. In this paper, we highlight a trade-off between explanation's stability and adherence, namely how much it resembles the ML model. Exploiting our innovative discovery, we propose a framework to maximise stability, while retaining a predefined level of adherence. OptiLIME provides freedom to choose the best adherence-stability trade-off level and more importantly, it clearly highlights the mathematical properties of the retrieved explanation. As a result, the practitioner is provided with tools to decide whether the explanation is reliable, according to the problem at hand. We extensively test OptiLIME on a toy dataset - to present visually the geometrical findings - and a medical dataset. In the latter, we show how the method comes up with meaningful explanations both from a medical and mathematical standpoint.