Search for: All records

The MRI-derived brain network serves as a pivotal instrument in elucidating both the structural and functional aspects of the brain, encompassing the ramifications of diseases and developmental processes. However, prevailing methodologies, often focusing on synchronous BOLD signals from functional MRI (fMRI), may not capture directional influences among brain regions and rarely tackle temporal functional dynamics. In this study, we first construct the brain-effective network via the dynamic causal model. Subsequently, we introduce an interpretable graph learning framework termed Spatio-Temporal Embedding ODE (STE-ODE). This framework incorporates specifically designed directed node embedding layers, aiming at capturing the dynamic interplay between structural and effective networks via an ordinary differential equation (ODE) model, which characterizes spatial-temporal brain dynamics. Our framework is validated on several clinical phenotype prediction tasks using two independent publicly available datasets (HCP and OASIS). The experimental results clearly demonstrate the advantages of our model compared to several state-of-the-art methods. 

Amphiphilic complexes with luminescent rare earth metal ions suitable for Lanmuir-Blodgett (LB) deposition have been synthesized. LB monolayers with closely packed Eu complexes deposited directly on silver demonstrate significant far-field emission in contrast to the theoretical predictions of full quenching. Angular radiation and polarization patterns of the electric and magnetic dipole emission of Eu³⁺point to a high excitation efficiency of surface plasmon polaritons. Different luminescent behavior of closely packed emitters in comparison to diluted systems is tentatively attributed to the collective state of emitters in LB layers formed via near-field coupling with surface plasmons. 

Abstract A photochemical model of photosynthetic electron transport (PET) is needed to integrate photophysics, photochemistry, and biochemistry to determine redox conditions of electron carriers and enzymes for plant stress assessment and mechanistically link sun‐induced chlorophyll fluorescence to carbon assimilation for remotely sensing photosynthesis. Towards this goal, we derived photochemical equations governing the states and redox reactions of complexes and electron carriers along the PET chain. These equations allow the redox conditions of the mobile plastoquinone pool and the cytochrome b₆f complex (Cyt) to be inferred with typical fluorometry. The equations agreed well with fluorometry measurements from diverse C₃/C₄species across environments in the relationship between the PET rate and fraction of open photosystem II reaction centres. We found the oxidation of plastoquinol by Cyt is the bottleneck of PET, and genetically improving the oxidation of plastoquinol by Cyt may enhance the efficiency of PET and photosynthesis across species. Redox reactions and photochemical and biochemical interactions are highly redundant in their complex controls of PET. Although individual reaction rate constants cannot be resolved, they appear in parameter groups which can be collectively inferred with fluorometry measurements for broad applications. The new photochemical model developed enables advances in different fronts of photosynthesis research. 

Abstract We have studied effects of metal–dielectric substrates on photopolymerization of [2,2ʹ-Bi-1H-indene]-1,1ʹ-dione-3,3ʹ-diyl diheptanoate (BITh) monomer. We synthetized BITh and spin-coated it onto a variety of dielectric, metallic, and metal–dielectric substrates. The films were exposed to radiation of a UV–Visible Xe lamp, causing photo-polymerization of monomer molecules. The magnitude and the rate of the photo-polymerization were monitored by measuring the strength of the ~ 480 nm absorption band, which existed in the monomer but not in the polymer. Expectedly, the rate of photo-polymerization changed nearly linearly with the change of the pumping intensity. In contrast with our early study of photo-degradation of semiconducting polymer P3HT, the rate of photo-polymerization of BITh is getting modestly higher if the monomer film is deposited on top of silver separated from the monomer by a thin insulating MgF 2 layer preventing a charge transfer. This effect is partly due to a constructive interference of the incident and reflected light waves, as well as known in the literature effects of metal/dielectric substrates on a variety of spectroscopic and energy transfer parameters. At the same time, the rate of photopolymerization is getting threefold larger if monomer is deposited on Ag film directly and charge transfer is allowed. Finally, Au substrates cause modest (~ 50%) enhancement of both monomer film absorption and the rate of photo-polymerization. 

The core of self-supervised learning for pre-training language models includes pre-training task design as well as appropriate data augmentation. Most data augmentations in language model pre-training are context-independent. A seminal contextualized augmentation was recently proposed in ELECTRA and achieved state-of-the-art performance by introducing an auxiliary generation network (generator) to produce contextualized data augmentation for the training of a main discrimination network (discriminator). This design, however, introduces extra computation cost of the genera- tor and a need to adjust the relative capability between the generator and the discriminator. In this paper, we propose a self-augmentation strategy (SAS) where a single network is utilized for both regular pre-training and contextualized data augmentation for the training in later epochs. Essentially, this strategy eliminates a separate generator and uses the single network to jointly conduct two pre-training tasks with MLM (Masked Language Modeling) and RTD (Replaced Token Detection) heads. It avoids the challenge to search for an appropriate size of the generator, which is critical to the performance as evidenced in ELECTRA and its subsequent variant models. In addition, SAS is a general strategy that can be seamlessly combined with many new techniques emerging recently or in the future, such as the disentangled attention mechanism from DeBERTa. Our experiments show that SAS outperforms ELECTRA and other state-of-the-art models in the GLUE tasks with similar or less computation cost. 

Artificial Intelligence (AI) brings advancements to support pathologists in navigating high-resolution tumor images to search for pathology patterns of interest. However, existing AI-assisted tools have not realized the promised potential due to a lack of insight into pathology and HCI considerations for pathologists’ navigation workflows in practice. We first conducted a formative study with six medical professionals in pathology to capture their navigation strategies. By incorporating our observations along with the pathologists’ domain knowledge, we designed NaviPath — a human-AI collaborative navigation system. An evaluation study with 15 medical professionals in pathology indicated that: (i) compared to the manual navigation, participants saw more than twice the number of pathological patterns in unit time with NaviPath, and (ii) participants achieved higher precision and recall against the AI and the manual navigation on average. Further qualitative analysis revealed that participants’ navigation was more consistent with NaviPath, which can improve the examination quality.