ScienceDaily: Computers & Math News

Researchers have created a taxonomy and outlined steps that developers can take to design features in machine-learning models that are easier for decision-makers to understand.

Current AIs are very accurate but inflexible at image recognition. Exactly why this is remains a mystery. Researchers have developed a method called 'Raw Zero-Shot' to assess how neural networks handle elements unknown to them. The results have the potential to help researchers identify the common features that make neural networks 'non-robust,' and develop methods to make AIs more reliable.

Computer scientists develop Emu, an algorithm that uses long reads of genomes to identify the species of bacteria in a community. The program could simplify sorting harmful from helpful bacteria in microbiomes like those in the gut or in agriculture and the environment.

The energy-levels of semiconductors can be rearranged by coupling between light particles (photons) and excited electrons so that a formerly dark material becomes optically active. In their experiments, the researchers succeeded in manipulating the energy-level structure in an ultra-thin sample of the semiconductor tungsten diselenide. This material, which normally has a low luminescence yield, began to shine, the team reports.