Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: information science – Page 236

Balloon shaping isn’t just for kids anymore. A team of researchers from the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) has designed materials that can control and mold a balloon into pre-programmed shapes. The system uses kirigami sheets—thin sheets of material with periodic cuts—embedded into an inflatable device. As the balloon expands, the cuts in the kirigami sheet guide the growth, permitting expansion in some places and constricting it in others. The researchers were able to control the expansion not only globally to make large-scale shapes, but locally to generate small features.

The team also developed an inverse design strategy, an algorithm that finds the optimum design for the kirigami inflatable device that will mimic a target shape upon inflation.

“This work provides a new platform for shape-morphing devices that could support the design of innovative medical tools, actuators and reconfigurable structures,” said Katia Bertoldi, the William and Ami Kuan Danoff Professor of Applied Mechanics at SEAS and senior author of the study.

Read more | >

Text is backward. Clocks run counterclockwise. Cars drive on the wrong side of the road. Right hands become left hands.

Intrigued by how reflection changes images in subtle and not-so-subtle ways, a team of Cornell researchers used artificial intelligence to investigate what sets originals apart from their reflections. Their algorithms learned to pick up on unexpected clues such as hair parts, gaze direction and, surprisingly, beards – findings with implications for training machine learning models and detecting faked images.

Read more | >

Scientists suggest a desktop quantum computer based on nuclear magnetic resonance (NMR) could soon be on its way to a classroom near you. Although the device might not be suited to handle large quantum applications, the makers say it could help students learn about quantum computing.

SpinQ Chief Scientist Prof. Bei Zeng from University of Guelph, announced the SpinQ Gemini, a two-qubit desktop quantum computer, at the industry session of the Quantum Information Processing (QIP2020) conference, which is held recently in Shenzhen, China. It is the first time that a desktop quantum computer is commercially available, according to the researchers.

SpinQ Gemini is built by the state-of-the-art technology of permanent magnets, providing 1T magnetic field, running at room temperature, and maintenance free. It demonstrates quantum algorithms such as Deutsch’s algorithm and Grover’s algorithm for teaching quantum computing to university and high school students, also provides advanced models for quantum circuit design and control sequence design for researchers.

Read more | >

However, the situation has been improving as Chinese tech giants including e-commerce company Alibaba, search engine Baidu, on-demand delivery company Meituan Dianping, ride-hailing operator Didi Chuxing and smartphone maker Xiaomi now offer more affordable health care plans via mutual aid platforms, which operate as a collective claim-sharing mechanism.

China’s online mutual aid platforms are disrupting old school insurance companies by leveraging big data and internet finance technologies to offer low cost medical coverage.

Read more | >

Updated mathematical techniques that can distinguish between two types of ‘non-Gaussian curve’ could make it easier for researchers to study the nature of quantum entanglement.

Quantum entanglement is perhaps one of the most intriguing phenomena known to physics. It describes how the fates of multiple particles can become entwined, even when separated by vast distances. Importantly, the probability distributions needed to define the quantum states of these particles deviate from the bell-shaped, or ‘Gaussian’ curves which underly many natural processes. Non-Gaussian curves don’t apply to quantum systems alone, however. They can also be composed of mixtures of regular Gaussian curves, producing difficulties for physicists studying quantum entanglement. In new research published in EPJ D, Shao-Hua Xiang and colleagues at Huaihua University in China propose a solution to this problem. They suggest an updated set of equations that allows physicists to easily check whether or not a non-Gaussian state is genuinely quantum.

As physicists make more discoveries about the nature of quantum entanglement, they are rapidly making progress towards advanced applications in the fields of quantum communication and computation. The approach taken in this study could prove to speed up the pace of these advances. Xiang and colleagues acknowledge that while all previous efforts to distinguish between both types of non-Gaussian curve have had some success, their choices of Gaussian curves as a starting point have so far meant that no one approach has yet proven to be completely effective. Based on the argument that there can’t be any truly reliable Gaussian reference for any genuinely quantum non-Gaussian state, the researchers present a new theoretical framework.

Read more | >

1. AI-optimized manufacturing

Paper and pencil tracking, luck, significant global travel and opaque supply chains are part of today’s status quo, resulting in large amounts of wasted energy, materials and time. Accelerated in part by the long-term shutdown of international and regional travel by COVID-19, companies that design and build products will rapidly adopt cloud-based technologies to aggregate, intelligently transform, and contextually present product and process data from manufacturing lines throughout their supply chains. By 2025, this ubiquitous stream of data and the intelligent algorithms crunching it will enable manufacturing lines to continuously optimize towards higher levels of output and product quality – reducing overall waste in manufacturing by up to 50%. As a result, we will enjoy higher quality products, produced faster, at lower cost to our pocketbooks and the environment.

Anna-Katrina Shedletsky, CEO and Founder of Instrumental.

Read more | >

Pagaya, an AI-driven institutional asset manager that focuses on fixed income and consumer credit markets, today announced it raised $102 million in equity financing. CEO Gal Krubiner said the infusion will enable Pagaya to grow its data science team, accelerate R&D, and continue its pursuit of new asset classes including real estate, auto loans, mortgages, and corporate credit.

Pagaya applies machine intelligence to securitization — the conversion of an asset (usually a loan) into marketable securities (e.g., mortgage-backed securities) that are sold to other investors — and loan collateralization. It eschews the traditional method of securitizing pools of previously assembled asset-backed securities (ABS) for a more bespoke approach, employing algorithms to compile discretionary funds for institutional investors such as pension funds, insurance companies, and banks. Pagaya selects and buys individual loans by analyzing emerging alternative asset classes, after which it assesses their risk and draws on “millions” of signals to predict their returns.

Pagaya’s data scientists can build algorithms to track activities, such as auto loans made to residents in cities and even specific neighborhoods, for instance. The company is only limited by the amount of data publicly available; on average, Pagaya looks at decades of information on borrowers and evaluates thousands of variables.

Read more | >

For instance, suppose a neural network has labeled the image of a skin mole as cancerous. Is it because it found malignant patterns in the mole or is it because of irrelevant elements such as image lighting, camera type, or the presence of some other artifact in the image, such as pen markings or rulers?

Researchers have developed various interpretability techniques that help investigate decisions made by various machine learning algorithms. But these methods are not enough to address AI’s explainability problem and create trust in deep learning models, argues Daniel Elton, a scientist who researches the applications of artificial intelligence in medical imaging.

Elton discusses why we need to shift from techniques that interpret AI decisions to AI models that can explain their decisions by themselves as humans do. His paper, “Self-explaining AI as an alternative to interpretable AI,” recently published in the arXiv preprint server, expands on this idea.

Read more | >