Toggle light / dark theme

This type of molecular collaboration has inspired scientists for nearly a century. Here, oxygen is the effector. It flips a protein switch, helping proteins better carry oxygen through the body. In other words, it may be possible to optimize protein functions with an alternative effector drug.

The problem? The original inspiration is wonky. Sometimes hemoglobin proteins carry oxygen. Other times they don’t. In 1965, a French and American collaboration found out why. Each protein alternates between two three-dimensional shapes—one that carries oxygen and another that doesn’t. The shapes can’t coexist in the assembled protein to carry oxygen: It’s all-or-none, depending on the presence and amount of the effector.

The new study built on these lessons to guide their AI-designed proteins.

Could food delivery robots with zero carbon emissions influence a customer’s decision to buy food using them instead of robot vehicles that emit carbon into the atmosphere? This is what a recent study published in the International Journal of Hospitality Management hopes to address as a tea of researchers from Washington State University (WSU) investigated how a customer’s knowledge of an automatic delivery robot’s (ADR) environment impact influences their choice regarding which type of robot they want delivering their food. This study holds the potential to help scientists, environmental conservationists, and the public better understand the benefits of eco-friendly delivery robots for both the short and long term.

“Much of the marketing focus has been on the functionality and the convenience of these automatic delivery robots, which is really important, but it would enhance these efforts to promote their green aspects as well,” said Jennifer Han, who is a doctoral student in WSU’s Carson College of Business and lead author of the study.

For the study, the researchers used the Amazon crowdsourcing platform, MTurk, to conduct an online survey comprised of 418 adults who were instructed to watch videos about ADRs followed by a questionnaire regarding the environmental impact and the risk of using ADRs for their food delivery service. In the end, the team discovered a connection between participants who found ADRs were less risky and wanted an eco-friendly ADR compared to participants who thought ADRs were riskier but weren’t concerned about the environmental consequences.

Inspired by biology we 1) get adversarial robustness + interpretability for free, 2) turn classifiers into generators & 3) design attacks on vLLMs.

Stanislav Fort, Balaji Lakshminarayanan August 2024 https://www.arxiv.org/abs/2408.


Adversarial examples pose a significant challenge to the robustness, reliability and alignment of deep neural networks. We propose a novel, easy-to-use approach to achieving high-quality representations that lead to adversarial robustness through the use of multi-resolution input representations and dynamic self-ensembling of intermediate layer predictions. We demonstrate that intermediate layer predictions exhibit inherent robustness to adversarial attacks crafted to fool the full classifier, and propose a robust aggregation mechanism based on Vickrey auction that we call \textit{CrossMax} to dynamically ensemble them.

Identifying one faulty turbine in a wind farm, which can involve looking at hundreds of signals and millions of data points, is akin to finding a needle in a haystack.

Engineers often streamline this complex problem using deep-learning models that can detect anomalies in measurements taken repeatedly over time by each turbine, known as time-series data.

But with hundreds of recording dozens of signals each hour, training a deep-learning model to analyze time-series data is costly and cumbersome. This is compounded by the fact that the model may need to be retrained after deployment, and wind farm operators may lack the necessary machine-learning expertise.