Lifeboat Foundation

As artificial intelligence (AI) becomes increasingly used for critical applications such as diagnosing and treating diseases, predictions and results regarding medical care that practitioners and patients can trust will require more reliable deep learning models.

In a recent preprint (available through Cornell University’s open access website arXiv), a team led by a Lawrence Livermore National Laboratory (LLNL) computer scientist proposes a novel deep learning approach aimed at improving the reliability of classifier models designed for predicting disease types from diagnostic images, with an additional goal of enabling interpretability by a medical expert without sacrificing accuracy. The approach uses a concept called confidence calibration, which systematically adjusts the model’s predictions to match the human expert’s expectations in the real world.

“Reliability is an important yardstick as AI becomes more commonly used in high-risk applications, where there are real adverse consequences when something goes wrong,” explained lead author and LLNL computational scientist Jay Thiagarajan. “You need a systematic indication of how reliable the model can be in the real setting it will be applied in. If something as simple as changing the diversity of the population can break your system, you need to know that, rather than deploy it and then find out.”

A picture may be worth a thousand words, but apparently one image is worth potentially thousands of headaches for Android users recently.

The noted tech information leaker Ice Universe this weekend posted a warning about an image that if set as wallpaper will soft-brick Samsung and Google Pixel phones. Soft-bricking triggers Android devices to continuously loop an action or freeze the unit. This generally requires a factory reset.

The image, a seemingly innocuous sunset (or dawn) sky above placid waters, may be viewed without harm. But if loaded as wallpaper, the phone will crash.

More than three-quarters of the baryonic content of the Universe resides in a highly diffuse state that is difficult to detect, with only a small fraction directly observed in galaxies and galaxy clusters^1,2. Censuses of the nearby Universe have used absorption line spectroscopy^3,4 to observe the ‘invisible’ baryons, but these measurements rely on large and uncertain corrections and are insensitive to most of the Universe’s volume and probably most of its mass. In particular, quasar spectroscopy is sensitive either to the very small amounts of hydrogen that exist in the atomic state, or to highly ionized and enriched gas^4,5,6 in denser regions near galaxies⁷. Other techniques to observe these invisible baryons also have limitations; Sunyaev–Zel’dovich analyses^8,9 can provide evidence from gas within filamentary structures, and studies of X-ray emission are most sensitive to gas near galaxy clusters^9,10. Here we report a measurement of the baryon content of the Universe using the dispersion of a sample of localized fast radio bursts; this technique determines the electron column density along each line of sight and accounts for every ionized baryon^11,12,13. We augment the sample of reported arcsecond-localized^{14,15,16,17,18} fast radio bursts with four new localizations in host galaxies that have measured redshifts of 0.291, 0.118, 0.378 and 0.522. This completes a sample sufficiently large to account for dispersion variations along the lines of sight and in the host-galaxy environments¹¹, and we derive a cosmic baryon density of \({\varOmega }_{{\rm{b}}}={0.051}_{-0.025}^{+0.021}{h}_{70}^{-1}\) (95 per cent confidence; h₇₀ = H₀/(70 km s⁻¹ Mpc⁻¹) and H₀ is Hubble’s constant). This independent measurement is consistent with values derived from the cosmic microwave background and from Big Bang nucleosynthesis^19,20.

Researchers in Italy have melded the emerging science of convolutional neural networks (CNNs) with deep learning — a discipline within artificial intelligence — to achieve a system of market forecasting with the potential for greater gains and fewer losses than previous attempts to use AI methods to manage stock portfolios. The team, led by Prof. Silvio Barra at the University of Cagliari, published their findings on IEEE/CAA Journal of Automatica Sinica.

The University of Cagliari-based team set out to create an AI-managed “buy and hold” (B&H) strategy — a system of deciding whether to take one of three possible actions — a long action (buying a stock and selling it before the market closes), a short action (selling a stock, then buying it back before the market closes), and a hold (deciding not to invest in a stock that day). At the heart of their proposed system is an automated cycle of analyzing layered images generated from current and past market data. Older B&H systems based their decisions on machine learning, a discipline that leans heavily on predictions based on past performance.

By letting their proposed network analyze current data layered over past data, they are taking market forecasting a step further, allowing for a type of learning that more closely mirrors the intuition of a seasoned investor rather than a robot. Their proposed network can adjust its buy/sell thresholds based on what is happening both in the present moment and the past. Taking into account present-day factors increases the yield over both random guessing and trading algorithms not capable of real-time learning.

Innovative vaccine uses another virus to present two genes from the novel coronavirus.

A duo of preclinical studies recently demonstrated a new way to ferry medicines past the blood-brain barrier. And other research is on the way.

Discovery Sheds New Light on Famous Einstein Ring

Social distance science made possible with public W. M. Keck Observatory and NASA archive data.

Determined to find a needle in a cosmic haystack, a pair of astronomers time traveled through archives of old data from W. M. Keck Observatory on Mauankea in Hawaii and old X-ray data from NASA’s Chandra X-ray Observatory to unlock a mystery surrounding a bright, lensed, heavily obscured quasar.

The thermoelectric generator harnesses the flow of heat between two surfaces — one exposed to the cold sky at night. It could be the nocturnal cousin of solar power, lighting the lives of the 1.7 billion people worldwide living with an unreliable electricity connection.

If a drop of creamer falls from a spoon into a swirling cup of coffee, the whirlpool drags the drop into rotation. But what would happen if the coffee had no friction—no way to pull the drop into a synchronized spin?