187037 research outputs found
Sort by
Preserve or Modify? Context-Aware Evaluation for Balancing Preservation and Modification in Text-Guided Image Editing
No full textThe development of vision-language and generative models has significantly advanced text-guided image editing, which seeks the preservation of core elements in the source image while implementing modifications based on the target text. However, existing metrics have a context-blindness problem, indiscriminately applying the same evaluation criteria on completely different pairs of source image and target text, biasing towards either modification or preservation. Directional CLIP similarity, the only metric that considers both source image and target text, is also biased towards modification aspects and attends to irrelevant editing regions of the image. We propose AugCLIP, a context-aware metric that adaptively coordinates preservation and modification aspects, depending on the specific context of a given source image and target text. This is done by deriving the CLIP representation of an ideally edited image, that preserves the source image with necessary modifications to align with target text. More specifically, using a multi-modal large language model, AugCLIP augments the textual descriptions of the source and target, then calculates a modification vector through a hyperplane that separates source and target attributes in CLIP space. Extensive experiments on five benchmark datasets, encompassing a diverse range of editing scenarios, show that AugCLIP aligns remarkably well with human evaluation standards, outperforming existing metrics. The code is available at https://github.com/augclip/augclip_eval
Autonomous chemo-metabolic construction of anisotropic cell-in-shell nanobiohybrids in enzyme-powered cell microrobots
No full textLiving organisms use intricate strategies to adapt and survive in response to potentially lethal environment changes. Inspired by cryptobiosis in nature, researchers have pioneered approaches to create cell-in-shell nanobiohybrids, aiming to endow cells with enhanced protection and exogenous functions. Yet, these methods still lack the biological autonomy intrinsic to natural cellular responses. Here, we present an innovative chemo-metabolically coupled strategy for the autonomous construction of cell-in-shell structures in cell growth medium. Our system harnesses ethanol fermentation by Saccharomyces cerevisiae, chemically coupled with an enzymatic cascade involving alcohol oxidase and horseradish peroxidase, to drive the nanoshell formation of polydopamine. The integration of autonomous shell formation with cellular proliferation produces anisotropic cell-in-shell structures, which can serve as enzyme-powered cell microrobots, upon conjugation with urease. Our autonomous system enables the creation of cell-in-shell nanobiohybrids with dynamic and adaptive environmental interactions, paving the way for transformative applications in synthetic biology, such as artificial cells, as well as advancements in cell-based therapies.
Predicting Mobile Payment Behavior Through Explainable Machine Learning and Application Usage Analysis
No full textIn the increasingly competitive mobile ecosystem, understanding user behavior is essential to improve targeted sales and the effectiveness of advertising. With the widespread adoption of smartphones and the increasing variety of mobile applications, predicting user behavior has become more complex. This study presents a comprehensive framework for predicting mobile payment behavior by integrating demographic, situational, and behavioral factors, focusing on patterns in mobile application usage. To address the complexity of the data, we use a combination of machine-learning models, including extreme gradient boosting, light gradient boosting machine, and CatBoost, along with Shapley additive explanations (SHAP) to improve interpretability. An analysis of extensive panel data from Korean Android users reveals that incorporating application usage behavior in such models considerably improves the accuracy of mobile payment predictions. The study identifies key predictors of payment behavior, indicated by high Shapley values, such as using social networking services (e.g., KakaoTalk and Instagram), media applications (e.g., YouTube), and financial and membership applications (e.g., Toss and OK Cashbag). Moreover, the results of the SHAP force analysis reveal the individual session-level drivers of mobile purchases. These findings advance the literature on mobile payment prediction and offer practical insights for improving targeted marketing strategies by identifying key behavioral drivers of mobile transactions.
Highly stable two-level current fluctuation in complex oxide heterostructures
No full textTwo-level systems based on point defects in dielectric oxides offer promising entropy source for random number generators. The random telegraph noise (RTN) generated by the two-level systems is ideal for creating random bit-strings for advanced computing and cryptographic technologies. However, in classical oxide systems, RTN signals often suffer from instability due to undesired defect migration and metastable electronic states. Herein, we present a two-level quantum system based on SrRuO3/LaAlO3/Nb-doped SrTiO3 heterostructure, which incorporates two different types of point defects, oxygen vacancies and antisite Ti defects. Temporal electron localization at antisite defects alters the energy levels of nearby oxygen vacancies through instantaneous Coulomb interaction, resulting in two-level current fluctuation across the interface. The RTN-like current signals exhibit high stability at room temperature. We utilize the stable two-level fluctuations to generate random bit-strings and confirm their applicability in practical stochastic machine learning algorithms for image super-resolution. This study provides a guideline for designing reliable entropy sources by exploiting the complementary interactions between cation and anion point defects in oxide-based electronic systems, essential for hardware-based random number generators.
Nucleation-promoting and growth-limiting synthesis of disordered rock-salt Li-ion cathode materials
No full textDisordered rock-salt oxides and oxyfluorides are promising positive electrode materials for high-performance lithium-ion batteries free of nickel and cobalt. However, conventional synthesis methods rely on post-synthesis pulverization to achieve cycling-appropriate particle sizes, offering limited control over particle microstructure and crystallinity. This accelerates degradation and complicates secondary particle processing. Here we present a synthesis strategy that enhances nucleation while suppressing particle growth and agglomeration across various disordered rock-salt compositions, including lithium-manganese-titanium oxide, lithium-manganese-niobium oxide, and lithium-nickel-titanium oxide systems. Applied to Li1.2Mn0.4Ti0.4O2, this method yields highly crystalline, well-dispersed sub-200 nm particles that form homogeneous electrode films with stable cycling behavior. Tested in cells with lithium metal as the counter electrode, these electrodes deliver similar to 200 mAh/g with 85% capacity retention relative to the first cycle after 100 cycles (20 mA/g, 1.5-4.8 V), and an average discharge voltage loss of 4.8 mV per cycle, compared to 38.6% retention and 7.5 mV loss per cycle for electrodes derived from pulverized solid-state particles. This approach suggests a route to enhance the performance and durability of disordered rock-salt electrodes for sustainable lithium-ion batteries.
Deciphering oxygen vacancies and d-band structures as key descriptors for understanding electrocatalytic trend of glycerol oxidation reaction kinetics in alkaline media
No full textDespite the advantages of employing electrocatalytic glycerol oxidation reaction (EGOR) as the anode reaction for hydrogen production, the limited understanding of how electrocatalytic activity trends relate to the intrinsic properties of electrocatalysts leads to frequent trial and error in designing high-performance electrocatalyst materials. Herein, we report systematic studies to decipher the correlation between the kinetics of EGOR under alkaline conditions and the concentration of oxygen vacancy (Vo) and d-band filling (fd) of electrocatalysts using a perovskite-based model system. The modulation of transition metal species in the B-site of LaMO3 (M = transition metal species such as Mn, Fe, Co, and Ni) electrocatalysts allows the systematic control of the density of Vo and fd of electrocatalysts. Our results, based on the experimental and computational calculation analyses, show that LaNiO3 (LNO) with the highest contents of Vo and fd exhibits the most enhanced intrinsic kinetics of EGOR. This is because accelerated charge transfer between LNO and glycolate species by high contents of Vo and fd weakens the strength of C-C bond in glycolate. In particular, the contents of Vo and fd have a proportional relationship with the specific electrocatalytic activity of LaMO3, indicating that Vo and fd can be used as descriptors to predict the specific activity of electrocatalysts. This work highlights the importance of controlling Vo and fd of electrocatalysts to achieve high performance EGOR.
Cooperation and Competition Among Automated Vehicles in Highway Merge ScenarioA Proof-of-Concept Study
No full textThis study compares strategies for controlling the specific movements of automated vehicles by applying concepts of cooperation and competition in a highway merge scenario. To enhance understanding, each concept was first analyzed from a sociological perspective. Based on this analysis, cooperation in the context of automated vehicles was characterized by the key terms "common goal/benefit, shared ownership of resources, and interaction", while competition was characterized by "individual goal/benefit, personal ownership of resources, and individual actions". Reflecting these key terms, the objective of cooperation was to optimize system cost, whereas the objective of competition was to optimize individual cost. Simulations were conducted to implement the movements of automated vehicles, and a dynamic programming model was used to predict and optimize vehicle trajectories. The results were analyzed in terms of mobility and safety metrics. Vehicles in the cooperation scenario, due to system-level optimization, maintained speed distributions centered around 90 km/h in the Main segment and 80 km/h in the merge segment. Competitive vehicles showed broader distributions with speeds peaking at 85-90 km/h and 75-80 km/h, respectively. In terms of safety, the cooperative approach recorded 26 conflicts in the merge segment compared to 33 in the competitive scenario, while achieving higher time-to-collision (TTC) values and lower deceleration rates (4.3 m/s(2) vs. 5.7 m/s(2)). These results highlight the advantages of the cooperative approach in optimizing both mobility and safety.