A5102967365- Digital Holography and Microscopy
- Recommender Systems and Techniques
- Advanced Fluorescence Microscopy Techniques
- Optical measurement and interference techniques
- Microfluidic and Bio-sensing Technologies
- Electrowetting and Microfluidic Technologies
- Advanced Bandit Algorithms Research
- Optical Coherence Tomography Applications
- Topic Modeling
- Near-Field Optical Microscopy
- Text and Document Classification Technologies
- Sentiment Analysis and Opinion Mining
- Data Stream Mining Techniques
- Photonic and Optical Devices
- Image Processing Techniques and Applications
- Advanced Optical Sensing Technologies
- Advanced Graph Neural Networks
- Cell Image Analysis Techniques
- Microfluidic and Capillary Electrophoresis Applications
- Advanced Text Analysis Techniques
- Advanced Optical Imaging Technologies
- Web Data Mining and Analysis
- Information Retrieval and Search Behavior
- Big Data and Business Intelligence
- Advanced Image and Video Retrieval Techniques
Home Depot (United States)
2020-2024
Georgia Institute of Technology
2023
Jiangsu University
2023
Academy of Military Medical Sciences
2014
California Institute of Technology
2006-2012
University of the Basque Country
2010
Jiangsu Ocean University
2007
Portland State University
2005
Zhejiang University
2002
Low-cost and high-resolution on-chip microscopes are vital for reducing cost improving efficiency modern biomedicine bioscience. Despite the needs, conventional microscope design has proven difficult to miniaturize. Here, we report implementation application of two (approximately 0.9 microm first approximately 0.8 second), lensless, fully based on optofluidic microscopy (OFM) method. These systems abandon design, which requires expensive lenses large space magnify images, instead utilizes...
Sensitive optical signal detection can often be confounded by the presence of a significant background, and, as such, predetection background suppression is substantively important for weak detection. In this paper, we present novel structure design, termed surface-wave-enabled darkfield aperture (SWEDA), which directly incorporated onto sensors to accomplish suppression. This SWEDA consists central hole and set groove pattern that channels incident light via surface plasmon wave...
We report the implementation of a color-capable on-chip lensless microscope system, termed color optofluidic (color OFM), and demonstrate imaging double stained Caenorhabditis elegans with lacZ gene expression at light intensity about 10 mW/cm(2).
Session-based recommender systems (SBRSs) have shown superior performance over conventional methods. However, they show limited scalability on large-scale industrial datasets since most models learn one embedding per item. This leads to a large memory requirement (of storing vector item) and poor sparse sessions with cold-start or unpopular items. Using public dataset, we experimentally that state-of-the-art SBRSs low We propose M2TRec, Metadata-aware Multi-task Transformer model for...
We report the implementation of an image sensor chip, termed wavefront chip (WIS), that can measure both intensity/amplitude and phase front variations a light wave separately quantitatively. By monitoring tightly confined transmitted spots through circular aperture grid in high Fresnel number regime, we intensity with sampling density (11 microm) sensitivity (the normalized gradient measurement is 0.1 mrad under typical working condition). using WIS standard microscope, collect bright-field...
We have developed a new microscopy design that can achieve wide field-of-view (FOV) imaging and yet possesses resolution is comparable to conventional microscope. In our design, the sample illuminated by holographically projected light-spot grid. acquire images translating across grid detecting transmissions. built prototype system with an FOV of 6mm×5mm acquisition time 2.5s. The fundamentally limited spot size—our demonstrated average FWHM diameter was 0.74μm. demonstrate U.S. Air Force...
E-commerce applications rely heavily on session-based recommendation algorithms to improve the shopping experience of their customers. Recent progress in shows great promise. However, translating that promise real-world outcomes is a challenging task for several reasons, but mostly due large number and varying characteristics available models. In this paper, we discuss approach lessons learned from process identifying deploying successful algorithm leading e-commerce application...
Optofluidic microscopy (OFM) is a novel technique for low-cost, high-resolution on-chip imaging. In this paper we report the use of Fresnel zone plate (FZP) based projection in OFM as cost-effective and compact means projecting transmission through an OFM's aperture array onto sensor grid. We demonstrate approach by employing FZP (diameter = 255 microm, focal length 800 microm) that has been patterned glass slide to project from apertures 1 separation 10 CMOS sensor. are able resolve...
Session-based recommender systems (SBRSs) predict users' next interacted items based on their historical activities. While most SBRSs capture purchasing intentions locally within each session, capturing items' global information across different sessions is crucial in characterizing general properties. Previous works this cross-session by constructing graphs and incorporating neighbor information. However, incorporation cannot vary adaptively according to the unique intention of constructed...
We experimentally measure and theoretically model the light transmission characteristics of subwavelength apertures. The characterization consists translating a point source at varying vertical height lateral displacement from aperture measuring resulting transmission. define variation with as collection mode spread function (CPSF). This geometry is particularly relevant to based imaging devices enables determination their resolution. study shows that achieved resolutions degrade sample...
We report a quantitative differential interference contrast (DIC) microscope based on structured-aperture (SA) wavefront sensor. Unlike conventional DIC microscope, the SA-DIC can separate amplitude and phase gradient information of image wavefront, form intensity images sample with good resolution; our prototype achieved resolution ∼2 μm. Furthermore, due to nonpolarization nature we were able birefringent samples without artifacts.
We present a novel generalized model for the analysis of noise with known spectral density. This is particularly appropriate 1/f(a) distribution in homodyne interferometer. The reveals that, alpha>1, significantly impacts signal-to-noise ratio (SNR) integration times that near characteristic time, beyond which SNR will no longer improve increasing time. experimentally verify our theoretical findings set experiments employing quadrature optical coherence tomography (OCT) system, finding good...
In this letter, the authors present a novel quadrature interferometry method based on use of harmonically matched shallow grating pair. Unlike simple beam splitter or single grating, pair can confer nontrivial interference phase shift (other than 0° 180°) between output ports interferometer. Using as splitter/combiner, implement homodyne full field interferometer and demonstrate system’s capability to acquire amplitude images.
We present a novel on-chip method for quantitative two-dimensional differential phase imaging. This technique uses four circular holes (600 nm diameter, 1.2 mum spacing) arranged in 'plus' pattern that are fabricated layer of metal above complementary metal-oxide semiconductor (CMOS) imaging sensor. The interference the aperture shifts position with respect to incident light. By CMOS sensor, this measures amplitude and (1 degrees /microm sensitivity signal-to-noise ratio >/=16 dB) light...
Session-based recommender systems (SBRSs) predict users' next interacted items based on their historical activities. While most SBRSs capture purchasing intentions locally within each session, capturing items' global information across different sessions is crucial in characterizing general properties. Previous works this cross-session by constructing graphs and incorporating neighbor information. However, incorporation cannot vary adaptively according to the unique intention of constructed...
Over the past few years, consumer behavior has shifted from traditional in-store shopping to online shopping. For example, eCommerce sales have grown around 5% of total US in 2012 15.4% year 2023. This rapid growth created new challenges and vital requirements for intelligent information retrieval systems. Which lead primary motivations this workshop:
We demonstrate the use of shallow diffraction gratings for quadrature phase interferometry. A single grating-based Michelson interferometer yields only trivial (0(o) or 180(o) ) shift between different output ports. In comparison, a combination two parallel can be useful to achieve desired shifts (e.g., 90 (o) interferometry). show that at ports grating-pair based adjusted by shearing with respect each other. Two harmonically-related are used control modified interferometer. Our experimental...
Boosting is a commonly used technique to enhance the performance of set base models by combining them into strong ensemble model. Though widely adopted, boosting typically in supervised learning where data labeled accurately. However, weakly learning, most through weak and noisy sources, it remains nontrivial design effective approaches. In this work, we show that standard implementation convex combination learners can hardly work due presence labels. Instead, propose $\textit{LocalBoost}$,...
We demonstrate a novel high-resolution portable beam profiler based on slanted linear array of small apertures, termed hole (SHArP). The apertures are directly fabricated metal-coated CMOS imaging sensor. With single scan, the aperture can establish virtual grid sampling points for profiling. our prototype, we profiling Gaussian beams over an area 66.5 microm x with resolution 0.8 (compare pixel size 10 microm). be improved into range submicrometers by fabricating smaller apertures. good...
On e-commerce platforms, predicting if two products are compatible with each other is an important functionality to achieve trustworthy product recommendation and search experience for consumers. However, accurately compatibility difficult due the heterogeneous data lack of manually curated training data. We study problem discovering effective labeling rules that can enable weakly-supervised prediction. develop AMRule, a multi-view rule discovery framework (1) adaptively iteratively discover...
An emerging dilemma that faces practitioners in large scale online experimentation for e-commerce is whether to use Multi-Armed Bandit (MAB) algorithms testing or traditional A/B (A/B). This paper provides a comprehensive comparison between the two, from perspectives of confidence intervals, hypothesis test powers, and their relationships with traffic split sample size both theoretically numerically. We first make comparisons MAB tests terms conditions under which disjoint intervals occur,...
Alternative recommender systems are critical for ecommerce companies. They guide customers to explore a massive product catalog and assist find the right products among an overwhelming number of options. However, it is non-trivial task recommend alternative that fit customers’ needs. In this paper, we use both textual information (e.g. titles descriptions) customer behavior data products. Our results show coverage significantly improved in offline evaluations as well recall precision. The...