“A time crystal is perhaps the simplest example of a non-equilibrium phase of matter,” said Yao, UC Berkeley associate professor of physics. “The QuTech system is perfectly poised to explore other out-of-equilibrium phenomena including, for example, Floquet topological phases.” These results follow on the heels of another time crystal sighting, also involving Yao’s group, published in Science several months ago. There, researchers observed a so-called prethermal time crystal, where the subharmonic oscillations are stabilized via high-frequency driving. The experiments were performed in Monroe’s lab at the University of Maryland using a one-dimensional chain of trapped atomic ions, the same system that observed the first signatures of time crystalline dynamics over five years ago. Interestingly, unlike the many-body localized time crystal, which represents an innately quantum Floquet phase, prethermal time crystals can exist as either quantum or classical phases of matter. Many open questions remain. Are there practical applications for time crystals? Can dissipation help to extend a time crystal’s lifetimes? And, more generally, how and when do driven quantum systems equilibrate?
Credit risk models are used by financial organizations to assess the credit risk of potential borrowers. Based on the credit risk model validation, they decide whether or not to approve a loan as well as the loan’s interest rate. New means of estimating credit risk have emerged as technology has progressed, like credit risk modelling using R and Python. Using the most up-to-date analytics and big data techniques to model credit risk is one of them. Other variables, such as the growth of economies and the creation of various categories of credit risk, have had an impact on credit risk modelling. Machine learning enables more advanced modelling approaches like decision trees and neural networks to be used. This introduces nonlinearities into the model, allowing for the discovery of more complex connections between variables. We selected to employ an XGBoost model that was fed with features picked using the permutation significance technique. ML models, on the other hand, are frequently so complex that they are difficult to understand. We chose to combine XGBoost and logistic regression because interpretability is critical in a highly regulated industry like credit risk assessment.
Up until now, AIOps has mostly been used in the context of monitoring. As a buzzword, people tend to think of the term in relation to creating baselines for your data and then alerting to any deviations, connecting multiple sources of information to find the root cause for a problem. These are powerful use cases and are allowing businesses to find correlations that they might not have achieved without AI. For example, you might find that poor bandwidth in a specific region led to an increase in tickets from customers within that location, or that you have idle cloud resources that are costing you in storage or compute dollars behind the scenes, allowing you to make manual changes to optimize costs. However, in many ways, categorizing AIOps as just monitoring and detection doesn’t make it that different from the previous category of IT operations analytics, where companies would look at operational data, including logs and security feeds, and then aggregate these to make smarter decisions. To get more out of AIOps, companies need to move past simple monitoring use cases, and look toward management of the cloud with the help of automation.
Quantum computing “systems” are still in development, and as such the entire system paradigm is in flux. While the race to quantum supremacy amongst nations and companies is picking up pace, it’s still at a very early stage to call it a “competition.” There are only a few potential qubit technologies deemed practical, the programming environment is nascent with abstractions that have still not been fully developed, and there are relatively few (albeit extremely exciting) quantum algorithms known to scientists and practitioners. Part of the challenge is that it is very difficult and nearly impractical to simulate quantum applications and technology on classical computers -- doing so would imply that classical computers have themselves outperformed their quantum counterparts! Nevertheless, governments are pouring funding into this field to help push humanity to the next big era in computing. The past decade has shown an impressive gain in qubit technologies, quantum circuits and compilation techniques are being realized, and the progress is leading to even more (good) competition towards the realization of full fledged quantum computers.
You may be wondering what’s the point of training an ML model. Well, for different use cases, there are different purposes. But in general, the purpose of training an ML model is more or less about making predictions on things that they’ve never seen. The model is about how to make good predictions. The way to create a model is called training — using existing data to identify a proper way to make predictions. There are many different ways to build a model, such as K-nearest neighbors, SVC, random forest, and gradient boosting, just to name a few. For the purpose of the present tutorial showing you how to build an ML model using Google Colab, let’s just use a model that’s readily available in sklearn — the random forest classifier. One thing to note is that because we have only one dataset. To test the model’s performance, we’ll split the dataset into two parts, one for training and the other for testing. We can simply use the train_test_split method, as shown below. The training dataset has 142 records, while the test dataset has 36 records, approximately in a ratio of 4:1.
In a global marketplace, being tied to one cloud service can be either impossible to achieve or not tolerable to the business, so cloud vendor neutrality becomes important. Practicality is one driver for always being open to multiple cloud solutions. There’s also vendor strategy: you will get driven, hard, to opt for its version of key parts of the software stack; their cloud version of database, say, as Amazon doesn’t want you to move your Oracle apps onto Amazon, but the Amazon database products instead. In and of itself, that’s not a crazy or risky decision; there are advantages to doing that if you are an Amazon customer already, and there will probably be economies of scale. It might even be cheaper (although that isn’t always the case with cloud contracts). This may feel like an easier solution to adopt, but could be a real headache if you have committed to microservices and have all kinds of open source apps running all over the place (as you want to, as it’s horses for courses here), and the database at the back is an Amazon database.
Collaboration is one of three primary use cases for a metaverse in the enterprise right now, according to Forrester VP J.P. Gownder. Another primary use case is one championed by chip giant Nvidia -- simulations and digital twins. Huang announced Nvidia Omniverse Enterprise during his keynote address at the company’s GTC 2021 online AI conference this month and offered several use cases that focused on simulations and digital twins in industrial settings such as warehouses, plants, and factories. If you are an organization in an industry with expensive assets -- for instance oil and gas, manufacturing, or logistics -- it makes sense to have this use case on your radar, according to Gartner’s Nguyen. “That’s where augmented reality is benefiting enterprise right now,” he says. As an example, during his keynote address, Nvidia’s Huang showed a video of a virtual warehouse created with Nvidia Omniverse Enterprise enabling an organization to visualize the impact of optimized routing in an automated order picking scenario. That’s an example of a particular use case, but Omniverse itself is Nvidia’s platform to enable organizations to create their own simulations or virtual worlds.
The FBI says the misconfiguration involved the Law Enforcement Enterprise Portal, or LEEP, which allows state, local and federal agencies to share information, including sensitive documents. The portal also supports a Virtual Command Center, which allows law enforcement agencies to share real-time information about events such as shootings and child abductions. Although the abused email server is operated by the FBI, the bureau issued an updated statement Sunday noting that the server is not part of the bureau's corporate email service, and that no classified systems or personally identifiable information was compromised. "No actor was able to access or compromise any data or PII on the FBI's network," the FBI says. "Once we learned of the incident, we quickly remediated the software vulnerability, warned partners to disregard the fake emails, and confirmed the integrity of our networks." The hacker-crafted note, a copy of which has been released by Spamhaus, warned that data had potentially been exfiltrated.
Resilience may be misunderstood as the ability to bounce back instantly from difficulties or to roll with any manner of punches. Defining or encouraging mindless acceptance of workplace stressors is a recipe for burnout. “The problem is when leadership focuses on building a team’s resilience as a way to avoid addressing unnecessary causes of stress that are part of the organization’s culture,” says David R. Caruso, Ph.D., author of A Leader’s Guide to Solving Challenges with Emotional Intelligence. “We will not, nor should we, try to meditate our way out of a toxic culture. Buying everyone a yoga mat while failing to address sources of unnecessary stress is a problem.” Therefore, it’s importance that IT leaders understand the state of affairs within the IT organization and actively address issues and drivers of burnout. “Burnout and change fatigue – disengagement that comes from constant or poorly managed change – are very real risks for team members and organizations,” says Noelle Akins, leadership coach and founder of Akins & Associates. “Resilience is not just non-stop adaptability.”
Improving employee happiness helps businesses, too. Utilizing this AI technology allows companies to track what's called "psychological capital," a phrase coined by professor Fred Luthans. Helping employees improve their scores not only makes people happier, but also significantly increases productivity and profit for companies. But what about using AI to make us happier outside of work? Happiness technologies have their place in our personal lives as well. ... One problem that the business world has been working on for years is finding a reliable, discreet way to analyze the day-to-day activities of employees to increase productivity and engagement. Research clearly shows that happiness is correlated to better performance, so this trend of tracking and improving employee happiness has gained popularity. On the back of this trend, AI is being used for robotic process automation (RPA) and content intelligence.
Quote for the day:
"Leadership is a matter of having people look at you and gain confidence, seeing how you react. If you're in control, they're in control." -- Tom Landry