We investigated the effect of different augmentation scenarios on predicting (retro)synthesis of chemical compounds using SMILES representation. We showed that augmentation of not only input sequences but also, importantly, of the target data eliminated the effect of data memorization by neural networks and improved their generalization performance for prediction of new sequences. The Top-5 accuracy was 85.4% for the prediction of the largest fragment (thus identifying principal transformation for classical retro-synthesis) for USPTO-50k test dataset and was achieved by a combination of SMILES augmentation and beam search. The same approach also outperformed best published results for prediction of direct reactions from the USPTO-MIT test set. Our model achieved 90.4% Top-1 and 96.5% Top-5 accuracy for its most challenging mixed set and 97% Top-5 accuracy for the USPTO-MIT separated set. The appearance frequency of the most abundantly generated SMILES was well correlated with the prediction outcome and can be used as a measure of the quality of reaction prediction.

We show a hardness result for random smoothing to achieve certified adversarial robustness against attacks in the $\ell_p$ ball of radius $\epsilon$ when $p>2$. Although random smoothing has been well understood for the $\ell_2$ case using the Gaussian distribution, much remains unknown concerning the existence of a noise distribution that works for the case of $p>2$. This has been posed as an open problem by Cohen et al. (2019) and includes many significant paradigms such as the $\ell_\infty$ threat model. In this work, we show that any noise distribution $\mathcal{D}$ over $\mathbb{R}^d$ that provides $\ell_p$ robustness for all base classifiers with $p>2$ must satisfy $\mathbb{E}\eta_i^2=\Omega(d^{1-2/p}\epsilon^2(1-\delta)/\delta^2)$ for 99% of the features (pixels) of vector $\eta\sim\mathcal{D}$, where $\epsilon$ is the robust radius and $\delta$ is the score gap between the highest-scored class and the runner-up. Therefore, for high-dimensional images with pixel values bounded in $[0,255]$, the required noise will eventually dominate the useful information in the images, leading to trivial smoothed classifiers.

Tinder has posted a warning for its users telling them that coronavirus safety is'more important' than dating and urging them to wash their hands frequently. The dating app also encourages its singletons to carry hand sanitiser, avoid touching their face and'maintain social distance' when out in public. The warning says: 'Tinder is a great place to meet new people. While we want you to continue to have fun, protecting yourself from the coronavirus is more important'. It appears as a pop up while users are flipping between potential matches to warn of the dangers of the potentially deadly virus now called COVID-19. The pop-up warning also includes a link to the latest advice and information from the World Health Organisation (WHO) website.

Viral particles are colorized purple in this color-enhanced transmission electron micrograph from a COVID-19 patient in the United States. Computer modeling can help epidemiologists predict how and where the illness will move next. Viral particles are colorized purple in this color-enhanced transmission electron micrograph from a COVID-19 patient in the United States. Computer modeling can help epidemiologists predict how and where the illness will move next. Scientists who use math and computers to simulate the course of epidemics are taking on the new coronavirus to try to predict how this global outbreak might evolve and how best to tackle it.

When Regina Barzilay had a routine mammogram in her early 40s, the image showed a complex array of white splotches in her breast tissue. The marks could be normal, or they could be cancerous--even the best radiologists often struggle to tell the difference. Her doctors decided the spots were not immediately worrisome. In hindsight, she says, "I already had cancer, and they didn't see it." Over the next two years Barzilay underwent a second mammogram, a breast MRI and a biopsy, all of which continued to yield ambiguous or conflicting findings. Ultimately she was diagnosed with breast cancer in 2014, but the path to that diagnosis had been unbelievably frustrating. "How do you do three tests and get three different results?" she wondered.

The writing is on the wall. Every major global governmental agency is warning of the imminent tipping point towards catastrophic climate change, even the world's largest oil company Saudi Aramco is now talking about reaching peak oil within the next 20 years, and the International Energy Agency projects that it will happen in more like 10. Solar and wind are cheaper than ever, and large-scale solar mega-projects are quickly becoming the norm. It makes sense, then, that even the supermajor oil companies are diversifying their portfolios and investing in their own demise--also known as the renewable energy sector. Way back in July, 2017 Oilprice reported that France's Total S.A. was "leading the charge on renewables". At the time, Total's website boasted: "For Total, contributing to the development of renewable energies is as much a strategic choice as an industrial responsibility. We are doing our part to diversify the global energy mix by investing in renewables, with a strategic focus on solar energy and bioenergies."

Digital technologies have transformed every industry in the Canadian economy. They have the power to help address some of the most challenging problems, from producing faster healthcare diagnoses to making businesses more efficient. They can improve quality of life in communities from anywhere in the world. Today, two Cape Breton technology companies announced new projects that will create well-paying jobs in emerging technology fields and generate sustainable economic growth in the Sydney area. Orenda Software Solutions Inc. is developing artificial intelligence and language understanding technology to measure social impacts and perform behavioural analysis.

China used locational and other data from hundreds of millions of smartphones to contain the spread of COVID-19, according to Chinese sources familiar with the program. In addition to draconian quarantine procedures, which kept more than 150 million Chinese in place at the February peak of the coronavirus epidemic, China used sophisticated computational methods on a scale never attempted in the West. With more than 80,000 cases registered, China reported only 126 new cases yesterday, compared to 851 in South Korea and 835 in Iran, out of a total of 1,969 new cases worldwide. Chinese sources emphasize that the artificial intelligence initiative supplemented basic public health measures, which centered on quarantines and aggressive efforts to convince Chinese citizens to change their behavior. Chinese government algorithms can estimate the probability that a given neighborhood or even an individual has exposure to COVID-19 by matching the location of smartphones to known locations of infected individuals or groups.

The spread of the COVID-19 coronavirus is a fluid situation changing by the day, and even by the hour. The growing worldwide public health emergency is threatening lives, but it's also impacting businesses and disrupting travel around the world. The OECD warns that coronavirus could cut global economic growth in half, and the Federal Reserve will cut the federal interest rates following the worst week for the stock market since 2008. Just how the COVID-19 coronavirus will affect the way we live and work is unclear because it's a novel disease spreading around the world for the first time, but it appears that AI may help fight the virus and its economic impact. A World Health Organization report released last month said that AI and big data are a key part of the response to the disease in China.

Security is always a concern for the enterprise, and learning new tactics to make it more effective is key. Machine learning (ML) and artificial intelligence (AI) play a big role for some companies, and our CIO jury agrees. Special report: Cybersecurity: Let's get tactical (free PDF) This ebook, based on the latest ZDNet / TechRepublic special feature, explores how organizations must adapt their security techniques, strengthen end-user training, and embrace new technologies like AI- and ML-powered defenses. When asked the question, "Are AI and ML a critical part of your cybersecurity plan this year," the 12-member CIO jury was definitely on the side of opting for innovation, with 67% saying they will be using these technologies to protect their company this year. Each of the remaining 33% of our panelists said that while they aren't using it yet, they are using AI and ML in other ways and learning more about it. Those weighing in on the'yes' side include Charles Eagan, CTO of BlackBerry.