56016 research outputs found
Sort by
An Overview of Machine Learning for Asset Management
No full textMachine learning has been widely used in the asset management industry to improve operations and make data-driven decisions. This article provides an overview of machine learning for asset management by presenting various machine learning models in the context of their applications, including general classification and regression, time-series forecasting, natural language processing, dimension reduction, reinforcement learning, data generation, recommendation, and clustering. Additionally, it highlights the challenges of implementing machine learning in asset management, such as data quality and quantity, interpretability, and fairness
Human-object interaction prediction in videos through gaze following
No full textUnderstanding the human-object interactions (HOIs) from a video is essential to fully comprehend a visual scene. This line of research has been addressed by detecting HOIs from images and lately from videos. However, the video-based HOI anticipation task in the third-person view remains understudied. In this paper, we design a framework to detect current HOIs and anticipate future HOIs in videos. We propose to leverage human gaze information since people often fixate on an object before interacting with it. These gaze features together with the scene contexts and the visual appearances of human-object pairs are fused through a spatio-temporal transformer. To evaluate the model in the HOI anticipation task in a multi-person scenario, we propose a set of person-wise multi-label metrics. Our model is trained and validated on the VidHOI dataset, which contains videos capturing daily life and is currently the largest video HOI dataset. Experimental results in the HOI detection task show that our approach improves the baseline by a great margin of 36.3% relatively. Moreover, we conduct an extensive ablation study to demonstrate the effectiveness of our modifications and extensions to the spatio-temporal transformer. Our code is publicly available on https://github.com/nizhf/hoi-prediction-gaze-transformer
Freestanding Oxide Membranes for Epitaxial Ferroelectric Heterojunctions
No full textSince facile routes to fabricate freestanding oxide membraneswerepreviously established, tremendous efforts have been made to furtherimprove their crystallinity, and fascinating physical properties havebeen also reported in heterointegrated freestanding membranes. Here,we demonstrate our synthetic recipe to manufacture highly crystallineperovskite SrRuO3 freestanding membranes using new infinite-layerperovskite SrCuO2 sacrificial layers. To accomplish this,SrRuO3/SrCuO2 bilayer thin films are epitaxiallygrown on SrTiO3 (001) substrates, and the topmost SrRuO3 layer is chemically exfoliated by etching the SrCuO2 template layer. The as-exfoliated SrRuO3 membranes aremechanically transferred to various nonoxide substrates for the subsequentBaTiO(3) film growth. Finally, freestanding heteroepitaxialjunctions of ferroelectric BaTiO3 and metallic SrRuO3 are realized, exhibiting robust ferroelectricity. Intriguingly,the enhancement of piezoelectric responses is identified in freestandingBaTiO(3)/SrRuO3 heterojunctions with mixed ferroelectricdomain states. Our approaches will offer more opportunities to developheteroepitaxial freestanding oxide membranes with high crystallinityand enhanced functionality
Real-time visualisation of ion exchange in molecularly confined spaces where electric double layers overlap
No full textIon interactions with interfaces and transport in confined spaces, where electric double layers overlap, are essential in many areas, ranging from crevice corrosion to understanding and creating nano-fluidic devices at the sub 10 nm scale. Tracking the spatial and temporal evolution of ion exchange, as well as local surface potentials, in such extreme confinement situations is both experimentally and theoretically challenging. Here, we track in real-time the transport processes of ionic species (LiClO4) confined between a negatively charged mica surface and an electrochemically modulated gold surface using a high-speed in situ sensing Surface Forces Apparatus. With millisecond temporal and sub-micrometer spatial resolution we capture the force and distance equilibration of ions in the confinement of D & AP; 2-3 nm in an overlapping electric double layer (EDL) during ion exchange. Our data indicate that an equilibrated ion concentration front progresses with a velocity of 100-200 & mu;m s(-1) into a confined nano-slit. This is in the same order of magnitude and in agreement with continuum estimates from diffusive mass transport calculations. We also compare the ion structuring using high resolution imaging, molecular dynamics simulations, and calculations based on a continuum model for the EDL. With this data we can predict the amount of ion exchange, as well as the force between the two surfaces due to overlapping EDLs, and critically discuss experimental and theoretical limitations and possibilities
Single-Molecule Approach for Biological Function of Schizosaccharomyces pombe AAA+ ATPase Abo1
No full text
Detection of Fiducial Marker With Neural Network Compression
No full textFiducial markers are used to localize camera positions and are widely employed in various fields where fast and highly accurate positioning is required, including AR (Augmented Reality), VR (Virtual Reality), PCB (Printed Circuit Board) design factories, and robot localization research. Over the past 20 years, many fiducial marker designs and detection algorithms have been proposed to improve detection rates, broaden the same marker family, or save computational resources. However, most of these algorithms work well in constrained environments, such as well-lit conditions, minimal motion blur, or no shadows. These limitations can be addressed by using learning-based methods, but they often suffer from high computational loads or the need for collecting training datasets. To overcome these limitations, we introduce a novel fiducial marker detection algorithm along with a neural network compression. By using a feature detection network with a simple circular-shape based fiducial marker, training datasets can be fully synthesized considering real-world noise without the effort of collecting and labeling datasets. Since many fiducial marker applications run on computationally constrained embedded systems, TD (Tensor Decomposition) and QAT (Quantization Aware Training) are applied to the neural network to reduce the number of parameters and improve the inference speed of the network. We demonstrate that our neural network compression approach preserves overall performance while reducing network parameters by 55.48% and accelerating inference speed by 569% on an NVIDIA Jetson Xavier NX. Furthermore, we validate our methods by testing them on real-world images taken by a flying drone
