10,000 publications from this institution
A nanoscale, visible-light, self-sensing stress probe would be highly desirable in a variety of biological, imaging, and materials engineering applications, especially a device that does not alter the mechanical properties of the material it seeks to probe. Here we present the CdSe–CdS tetrapod quantum dot, incorporated into polymer matrices via electrospinning, as an in situ luminescent stress probe for the mechanical properties of polymer fibers. The mechanooptical sensing performance is enhanced with increasing nanocrystal concentration while causing minimal change in the mechanical properties even up to 20 wt % incorporation. The tetrapod nanoprobe is elastic and recoverable and undergoes no permanent change in sensing ability even upon many cycles of loading to failure. Direct comparisons to side-by-side traditional mechanical tests further validate the tetrapod as a luminescent stress probe. The tetrapod fluorescence stress–strain curve shape matches well with uniaxial stress–strain curves measured mechanically at all filler concentrations reported.
Planning at a higher level of abstraction instead of low level torques improves the sample efficiency in reinforcement learning, and computational efficiency in classical planning. We propose a method to learn such hierarchical abstractions, or subroutines from egocentric video data of experts performing tasks. We learn a self-supervised inverse model on small amounts of random interaction data to pseudo-label the expert egocentric videos with agent actions. Visuomotor subroutines are acquired from these pseudo-labeled videos by learning a latent intent-conditioned policy that predicts the inferred pseudo-actions from the corresponding image observations. We demonstrate our proposed approach in context of navigation, and show that we can successfully learn consistent and diverse visuomotor subroutines from passive egocentric videos. We demonstrate the utility of our acquired visuomotor subroutines by using them as is for exploration, and as sub-policies in a hierarchical RL framework for reaching point goals and semantic goals. We also demonstrate behavior of our subroutines in the real world, by deploying them on a real robotic platform. Project website: this https URL.
No abstract is provided for this article.
Recent advances in instruction-following large language models (LLMs) have led to dramatic improvements in a range of NLP tasks. Unfortunately, we find that the same improved capabilities amplify the dual-use risks for malicious purposes of these models. Dual-use is difficult to prevent as instruction-following capabilities now enable standard attacks from computer security. The capabilities of these instruction-following LLMs provide strong economic incentives for dual-use by malicious actors. In particular, we show that instruction-following LLMs can produce targeted malicious content, including hate speech and scams, bypassing in-the-wild defenses implemented by LLM API vendors. Our analysis shows that this content can be generated economically and at cost likely lower than with human effort alone. Together, our findings suggest that LLMs will increasingly attract more sophisticated adversaries and attacks, and addressing these attacks may require new approaches to mitigations.
Evidence for the importance of sleep in the mood disorders has mushroomed over the past decade. Among adolescents and adults with a mood disorder, sleep disturbance is a risk factor for episodes, can contribute to relapse, has an adverse impact on emotion regulation, is critical for cognitive functioning, compromises health, and may contribute to substance use comorbidity and suicidality. This evidence has triggered a shift away from viewing sleep disturbance as an epiphenomenon, toward viewing sleep disturbance as an important but under-recognized mechanism in the multifactorial cause and maintenance of the mood disorders. Because the biology underpinning the sleep and circadian system is an open system, readily influenced by inputs from the environment, sleep in the mood disorders represents a unique and exciting domain for interdisciplinary research across behavioral, social, cognitive, and neurobiological levels of explanation. Together, the accumulated evidence has informed a range of novel, powerful, simple, and inexpensive treatments with potential for massive improvements to public health, including improving quality of life, reducing length and severity of episodes, and reducing the risk of subsequent episodes in the large number of individuals who suffer from mood disorders.
