When asked to copy a given sentence, T5 was able to “mindtext” at about 90 characters per minute (roughly 45 words by one estimate), “the highest typing rate that has yet been reported for any type of BCI,” the team wrote, and a twofold improvement over previous setups. His freestyle typing—answering questions—overall matched in performance, and met the average speed of thumb texting of his age group. “Willett and co-workers’ study begins to deliver on the promise of BCI technologies,” said Rajeswaran and Orsborn, not just for mind-texting, but also what comes next. The idea of tapping into machine learning algorithms is smart, yes, because the field is rapidly improving—and illustrating another solid link between neuroscience and AI. But perhaps more importantly, an algorithm’s performance relies on good data. Here, the team found that the time difference between writing letters, something rather complex, is what made the algorithm perform so well. In other words, for future BCIs, “it might be advantageous to decode complex behaviors rather than simple ones, particularly for classification tasks.”
These codes can invoke various actions on a smartphone device. Here lies the threat. While a QR code may appear as though it is designed to help us sign in to a Wi-Fi network or be part of an innocent marketing campaign, the intent of it may be entirely different, with threat actors architecting and deploying malicious codes in a variety of ways. They can be used to direct the user to a malicious URL for the purpose of phishing; force a call, thereby exposing the end user’s phone number to a scammer or a potentially expensive call centre; send a payment within seconds; obtain a user’s location; or draft an email or text and populate the recipient and subject lines. Additionally, they may introduce a compromised network on a device’s preferred network list and include a credential that enables the device to automatically connect to that network. Once connected, an attacker could launch further ‘Man-in-the-Middle’ attacks. Given the variety and seriousness of these potential threats, some key statistics released by MobileIron in September 2020 provide cause for alarm. ... At the same time, however, 71% of respondents said they could not “distinguish between a legitimate and malicious QR code”.
Understanding the importance of long-term security investment is one thing. Putting this into practice presents a new challenge entirely. Organizations can look to assign this task internally, but ultimately, IT personnel need to focus on business goals. At the same time, it is not always reasonable to expect an IT department to keep up with developments on topics ranging from software security to cryptography to hardware architecture. By relying on vendors for security agility, organizations can outsource the technologies required for long-term protection. Enterprise service management (ESM) is one such option that looks to ensure organizations are protected for a matter of years. As digital transformation accelerates, new technologies that enable the business tend to go out of date entirely too soon after deploying. Investing and enhancing the IT department to tackle the challenge can be one way forward, but ultimately businesses need their employees to focus on their own expertise. ESM enables the adoption of new technologies whilst allowing for organizations to move at their own pace.
Work that involves self-reflection is challenging, and when it comes to addressing diversity issues and one’s position within them, some IT leaders may be unsure about how to effectively approach these sensitive topics. But being an ally can set an example for the entire company. Allies in leadership positions can demonstrate to employees that conversations about diversity, equity, and inclusion (DEI) are encouraged, and that employees from traditionally marginalized groups have a safe environment to bring up concerns they may have about their own experiences in the workplace. “I think it’s about setting an example from the top, being willing to engage in those uncomfortable conversations. A really effective ally is willing to put some skin in the game, to put some of their privilege on the line, on behalf of someone who doesn’t have that privilege,” says Malcom Glenn, director of public affairs at Better.com and former head of global policy for accessibility and underserved communities at Uber. Conversations around racism, sexism, and bigotry, both within and outside of the workplace, can be very difficult.
As the CISO role evolves and more data is stored in clouds, DeFiore said her priorities right now are moving "back to basics" and knowing where the airline's data is, applying patches and working from a stance of least privilege. Also important is reducing the attack surface, she said, and "making sure we're only publishing things to the internet that need to be there and segmenting and making sure there's no opportunities for lateral movement" inside the network. Twitter's mission is to protect public conversations, and Sethi said that requires being able to recover quickly. She also said she thinks there will be an increase in the number of security vendors suffering breaches, "which is why I say think about who you partner with." Salem said he was impressed with "how well CISOs responded during the pandemic," and moved from a world in which they had a lot of control to very little--almost overnight. The lesson the security community has learned from that experience is to be agile, he said. Looking ahead, the CISO needs to continue becoming integrated into the day-to-day operations of the business so they can be better prepared, he said.
Forecasting based on historic trends has long been an essential part of conducting business. Whether using seasonal averages or industry knowledge, predicting the future was always a use-case with numerous applications. Data Scientists have a good mix of talents that line them up with the request to fit forecasting models, but they have not always had the educational background to effectively communicate their results in terms of economics and finance. ... As more fitted models start making their way to production, a knowledge gap in deploying and maintaining models emerges. A 100% accurate model that only lives on your machine is close to 0% useful. The ETL (extract, transform and load) and packaging of ML capabilities with requirements is a grey area for many current data scientists, and not always covered in training programs. Auto-ML capabilities are increasing the frequency that a useful model, designed to be consumed, will be thought of for production. In my opinion, this is an area of focus that will become the most lucrative in terms of jobs and compensation.
Data security concerns are top of mind for bank leaders. An important part of understanding the cloud is considering how an enterprise’s current infrastructure and capabilities may be limiting its ability to detect and address new risks and vulnerabilities—and how cloud technology can help. Security is different in the cloud because of the tools that are native to each cloud provider’s environment and the fact that cloud providers typically take responsibility for the security of the lower-level infrastructure layers. The shared security responsibility between cloud providers and the clients they host changes how organizations should anticipate and prepare for security risks. ... Cloud computing can help banks and financial services firms meet ever-evolving regulatory reporting requirements (e.g., Comprehensive Capital Analysis and Review, Solvency II) in multiple operating jurisdictions—a critically important capability in an industry where cross-border transactions are the norm. Cloud solutions can also help banks conduct intraday liquidity and risk calculations, and mine trade surveillance data to detect anti-money laundering and other fraud issues.
Globally, across all the markets surveyed by Ericsson where 5G commercial networks are available, an average 4% of consumers own a 5G smartphone and have a 5G subscription. In the UK, an overwhelming 97% of respondents are yet to embrace next-generation connectivity. This is partly due to the lack of clarity in the marketing of 5G, which has confused customers as to what the technology is and what it can offer. Heavy tech jargon and misinformation campaigns have in some cases even put users off entirely from planning to upgrade. In the UK, for example, the number of consumers intending to upgrade to 5G next year stands at 25% – down from 27% in 2019. For the few who have switched to 5G networks, however, the experience overall seems positive, with better levels of satisfaction recorded compared to users connected with 4G LTE. Perhaps as a reflection of 5G's capabilities, Wi-Fi usage among those who have upgraded is reducing. A quarter of respondents, said Ericsson, have either decreased or stopped using Wi-Fi after switching to 5G.
IT leaders putting new cloud and technology strategies in place to help improve their resiliency and availability must ensure that, once implemented, they don’t just sit back and hope that it works. Organisations need to know that the improvements to their business continuity plans (BCPs) will stand the test of disruption when further unexpected, and even significant planned-for, events occur further down the line. Stress-testing exercises are the best way of identifying gaps and faults in your BCP, before it is too late and remediating any problems becomes reactive rather than proactive. Organisations can do this by identifying which threats pose the biggest risk to the business – this is a vital step as every company and industry is different. Once these have been identified, each scenario can be placed in order of priority, by gauging exactly how much impact they could have on the business and how complex the plan needs to be to respond. When conducting these exercises, each team member should also be assigned a clear role. This might be as an active player in the BCP test, or it may be as an external party, for example an evaluator or observer, who can help spot any flaws and play a vital role in measuring its success.
A generative adversarial network (GAN) is a system that is composed of two deep-learning models: a generator which learns to create realistic data and a discriminator which learns to distinguish between real data and the generator's output. After training, often the generator is used alone, to simply produce data. NVIDIA has used GANs for several applications, including its Maxine platform for reducing video-conference bandwidth. In 2019, NVIDIA developed a GAN called StyleGAN that can produce photorealistic images of human faces and is used in the popular website This Person Does Not Exist. Last year, NVIDIA developed a variation of StyleGAN that can take as input the desired camera, texture, background, and other data, to produce customizable renderings of an image. Although GANs can produce an infinite number of unique high-quality images, most CV training algorithms also require that images be annotated with information about the objects in the image. ImageNet, one of the most popular CV datasets, famously employed tens of thousands of workers to label images using Amazon's Mechanical Turk.
Quote for the day:
"A leader is one who sees more than others see and who sees farther than others see and who sees before others see." -- Leroy Eimes