Search for: All records

The problem of few-shot graph classification targets at assigning class labels for graph samples, where only limited labeled graphs are provided for each class. To solve the problem brought by label scarcity, recent studies have proposed to adopt the prevalent few-shot learning framework to achieve fast adaptations to graph classes with limited labeled graphs. In particular, these studies typically propose to accumulate meta-knowledge across a large number of meta-training tasks, and then generalize such meta-knowledge to meta-test tasks sampled from a disjoint class set. Nevertheless, existing studies generally ignore the crucial task correlations among meta-training tasks and treat them independently. In fact, such task correlations can help promote the model generalization to meta-test tasks and result in better classification performance. On the other hand, it remains challenging to capture and utilize task correlations due to the complex components and interactions in meta-training tasks. To deal with this, we propose a novel few-shot graph classification framework FAITH to capture task correlations via learning a hierarchical task structure at different granularities. We further propose a task-specific classifier to incorporate the learned task correlations into the few-shot graph classification process. Moreover, we derive FAITH+, a variant of FAITH that can improve the sampling process for the hierarchical task structure. The extensive experiments on four prevalent graph datasets further demonstrate the superiority of FAITH and FAITH+ over other state-of-the-art baselines.

 

Traditional gender roles and gender stereotypes assume different life and career priorities among men and women. Meanwhile, the science profession is commonly considered to abide by a universalist ethos and a meritocracy that is independent of gender. We examined whether men and women scientists held different ideals about the good life and about good science. Furthermore, we investigated if those ideals of good life and of good science were linked in the minds of scientists; and if the linkages differed by gender. This study used a structural topic modeling approach to analyze the interview transcripts of 108 women and 92 men elite scientists who had received highly prestigious postdoctoral fellowships during the 1960s and1970s. In the open-ended interviews, the scientists were asked to describe their ideals of good life and of good science. Regarding the good life, we found that women scientists focused more on enjoying life and relationships and less on intellectual stimulation, relative to men scientists. For good science, women scientists focused more on empirical procedural accuracy and less on basic and fundamental breakthroughs, relative to men scientists. Moreover, we found that women scientists exhibited correlations between life and science ideals, whereas the two domains were completely separate for men scientists. In conclusion, a gendered system of life and science ideals existed even among this group of highly promising scientists.

 

Since the 14th Critical Assessment of Techniques for Protein Structure Prediction (CASP14), AlphaFold2 has become the standard method for protein tertiary structure prediction. One remaining challenge is to further improve its prediction. We developed a new version of the MULTICOM system to sample diverse multiple sequence alignments (MSAs) and structural templates to improve the input for AlphaFold2 to generate structural models. The models are then ranked by both the pairwise model similarity and AlphaFold2 self-reported model quality score. The top ranked models are refined by a novel structure alignment-based refinement method powered by Foldseek. Moreover, for a monomer target that is a subunit of a protein assembly (complex), MULTICOM integrates tertiary and quaternary structure predictions to account for tertiary structural changes induced by protein-protein interaction. The system participated in the tertiary structure prediction in 2022 CASP15 experiment. Our server predictor MULTICOM_refine ranked 3rd among 47 CASP15 server predictors and our human predictor MULTICOM ranked 7th among all 132 human and server predictors. The average GDT-TS score and TM-score of the first structural models that MULTICOM_refine predicted for 94 CASP15 domains are ~0.80 and ~0.92, 9.6% and 8.2% higher than ~0.73 and 0.85 of the standard AlphaFold2 predictor respectively.

 

Metalorganic chemical vapor deposition (MOCVD) is a promising technique for wafer-scale synthesis of MoS2 monolayers for 2D field-effect transistors (2D-FETs) and related devices. Epitaxial growth of MoS2 on sapphire provides films that are crystallographically well-oriented but typically contain low-angle grain boundaries (e.g., mirror twins), voids, and other defects depending on growth conditions and substrate characteristics. In this study, we investigate microstructure, optical properties, and field-effect characteristics of wafer-scale MoS2 monolayers grown by MOCVD on c-plane sapphire over a narrow window of growth temperatures (900–1000 °C). The density of low-angle grain boundaries in the MoS2 monolayer was found to decrease dramatically from 50% areal coverage for films grown at 900 °C to 5% at 1000 °C. This decrease in low-angle grain boundary density is correlated with an increase in the room-temperature photoluminescence intensity of A excitons and a decrease in the full-width-half maximum (FWHM) of the Raman A1g peak, which are typically indicative of a general reduction in defects in MoS2. However, the best transport properties (e.g., mean field-effect mobility mFE = 17.3 cm2/V s) were obtained in MoS2 monolayers grown at an intermediate temperature of 950 °C. It was found that as the growth temperature increased, small regions bound by high-angle boundaries begin to appear within the monolayer and increase in areal coverage, from ∼2% at 900 °C to ∼5% at 950 °C to ∼10% at 1000 °C. The growth temperature of 950 °C, therefore, provides an intermediate condition where the combined effects of low-angle and high-angle boundaries are minimized. The results of this study provide guidance on MOCVD growth and characterization that can be used to further optimize the performance of MoS2 2D-FETs.

 

Digital media are pervasive in the lives of young people and provide opportunities for them to learn about STEM. Multiple theories argue that the STEM media environment may shape how youth see a STEM career in their future. Yet, little is known about how pre-college digital media consumption may be related to students’ STEM career interest at the beginning of college. The wide variety of STEM media also raises the question of potentially different effects and pathways by media type. In this study, we collected a nationally representative sample of more than 15,000 students in their first year in U.S. colleges and universities. We asked about their career interests at the beginning of college and also asked them to retrospectively report their STEM media consumption during high school.

We found that watching STEM-related TV and online videos, as well as playing STEM-related video games during high school, were positively associated with students’ STEM career interests at the beginning of college. However, we also found that STEM media consumption did not impact directly on STEM career interest, but acted through two intermediaries: STEM identity (I and others see me as a STEM person) and three personal career outcome expectations: a high interest in self-development (enhancement and use of talents), and low interests in material status (money, fame, power) and in interpersonal relationships (helping, and working with, other people).

This study finds that STEM media have a significant effect in fostering STEM career interest, with most of the effect coming from STEM TV, STEM video viewing, and STEM video games. The effect is mediated mainly through students’ identity and, to a lesser extent, through personal values, such as self-development, material, and interpersonal relationship values. This study suggests that media communication should be mindful of how different platforms may deliver nuanced and varied messages of what STEM careers may afford and who can succeed in STEM.

 

Microelectronic devices can directly communicate with biology, as electronic information can be transmitted via redox reactions within biological systems. By engineering biology’s native redox networks, we enable electronic interrogation and control of biological systems at several hierarchical levels: proteins, cells, and cell consortia. First, electro-biofabrication facilitates on-device biological component assembly. Then, electrode-actuated redox data transmission and redox-linked synthetic biology allows programming of enzyme activity and closed-loop electrogenetic control of cellular function. Specifically, horseradish peroxidase is assembled onto interdigitated electrodes where electrode-generated hydrogen peroxide controls its activity.E. coli’s stress response regulon,oxyRS, is rewired to enable algorithm-based feedback control of gene expression, including an eCRISPR module that switches cell-cell quorum sensing communication from one autoinducer to another—creating an electronically controlled ‘bilingual’ cell. Then, these disparate redox-guided devices are wirelessly connected, enabling real-time communication and user-based control. We suggest these methodologies will help us to better understand and develop sophisticated control for biology.

 

Federated Learning (FL) enables multiple clients to collaboratively learn a machine learning model without exchanging their own local data. In this way, the server can exploit the computational power of all clients and train the model on a larger set of data samples among all clients. Although such a mechanism is proven to be effective in various fields, existing works generally assume that each client preserves sufficient data for training. In practice, however, certain clients can only contain a limited number of samples (i.e., few-shot samples). For example, the available photo data taken by a specific user with a new mobile device is relatively rare. In this scenario, existing FL efforts typically encounter a significant performance drop on these clients. Therefore, it is urgent to develop a few-shot model that can generalize to clients with limited data under the FL scenario. In this paper, we refer to this novel problem as federated few-shot learning. Nevertheless, the problem remains challenging due to two major reasons: the global data variance among clients (i.e., the difference in data distributions among clients) and the local data insufficiency in each client (i.e., the lack of adequate local data for training). To overcome these two challenges, we propose a novel federated few-shot learning framework with two separately updated models and dedicated training strategies to reduce the adverse impact of global data variance and local data insufficiency. Extensive experiments on four prevalent datasets that cover news articles and images validate the effectiveness of our framework compared with the state-of-the-art baselines. 

Autophagy is a cellular process with important functions that drive neurodegenerative diseases and cancers. Lysosomal hyperacidification is a hallmark of autophagy. Lysosomal pH is currently measured by fluorescent probes in cell culture, but existing methods do not allow for quantitative, transient or in vivo measurements. In the present study, we developed near-infrared optical nanosensors using organic color centers (covalent sp3 defects on carbon nanotubes) to measure autophagy-mediated endolysosomal hyperacidification in live cells and in vivo. The nanosensors localize to the lysosomes, where the emission band shifts in response to local pH, enabling spatial, dynamic and quantitative mapping of subtle changes in lysosomal pH. Using the sensor, we observed cellular and intratumoral hyperacidification on administration of mTORC1 and V-ATPase modulators, revealing that lysosomal acidification mirrors the dynamics of S6K dephosphorylation and LC3B lipidation while diverging from p62 degradation. This sensor enables the transient and in vivo monitoring of the autophagy–lysosomal pathway.