Delay is an important metric to understand and improve system performance. While existing approaches focus on aggregated delay statistics in pre-programmed granularity and provide results such as average and deviation, those approaches may not provide fine-grained delay measurement and thus may miss important delay characteristics. For example, delay anomaly, which is a critical system performance indicator, may not be captured by coarse-grained approaches. We propose a new measurement structure design called order preserving aggregator (OPA). Based on OPA, we can efficiently encode and recover the ordering and loss information by exploiting inherent data characteristics. We then propose a two-layer design to convey both ordering and time stamp, and efficiently derive per-packet delay/loss measurement. We evaluate our approach both analytically and experimentally. The results show that our approach can achieve per-packet delay measurement with an average of per-packet relative error at 2%, and an average of aggregated relative error at , while introducing additional communication overhead in the order of in terms of number of packets. While at a low data rate, the computation overhead of OPA is acceptable. Reducing the computation and communication overhead under high data rate, to make OPA more practical in real applications, will be our future direction.
Background & Aims: Cholangiocarcinoma (CCA) is a highly lethal malignancy originating from the biliary ducts. Current CCA diagnostic and prognostic assessments cannot satisfy the clinical requirement. Bile detection is rarely performed, and herein, we aim to estimate the clinical significance of bile liquid biopsy by assessing bile exosomal concentrations and components. Approach & Results: Exosomes in bile and sera from CCA, pancreatic cancer, and common bile duct stone were identified and quantified by transmission electronmicroscopy, nanoparticle tracking analysis, and nanoFCM. Exosomal components were assessed by liquid chromatography with tandem mass spectrometry and microRNA sequencing (miRNA-seq). Bile exosomal concentration in different diseases had no significant difference, but miR-182-5p and miR-183-5p were ectopically upregulated in CCA bile exosomes. High miR-182/183-5p in both CCA tissues and bile indicates a poor prognosis. Bile exosomal miR-182/183-5p is secreted by CCA cells and can be absorbed by biliary epithelium or CCA cells. With xenografts in humanized mice, we showed that bile exosomal miR-182/183-5p promotes CCA proliferation, invasion, and epithelial-mesenchymal transition (EMT) by targeting hydroxyprostaglandin dehydrogenase in CCA cells and mast cells (MCs), and increasing prostaglandin E2 generation, which stimulates PTGER1 and increases CCA stemness. In single-cell mRNA-seq, hydroxyprostaglandin dehydrogenase is predominantly expressed in MCs. miR-182/183-5p prompts MC to release VEGF-A release from MC by increasing VEGF-A expression, which facilitates angiogenesis. Conclusions: CCA cells secret exosomal miR-182/183-5p into bile, which targets hydroxyprostaglandin dehydrogenase in CCA cells and MCs and increases prostaglandin E2 and VEGF-A release. Prostaglandin E2 promotes stemness by activating PTGER1. Our results reveal a type of CCA self-driven progression dependent on bile exosomal miR-182/183-5p and MCs, which is a new interplay pattern of CCA and bile.
Neural architecture search automates neural network design and has achieved state-of-the-art results in many deep learning applications. While recent literature has focused on designing networks to maximize accuracy, little work has been conducted to understand the compatibility of architecture design spaces to varying hardware. In this paper, we analyze the neural blocks used to build Once-for-All (MobileNetV3), ProxylessNAS and ResNet families, in order to understand their predictive power and inference latency on various devices, including Huawei Kirin 9000 NPU, RTX 2080 Ti, AMD Threadripper 2990WX, and Samsung Note10. We introduce a methodology to quantify the friendliness of neural blocks to hardware and the impact of their placement in a macro network on overall network performance via only end-to-end measurements. Based on extensive profiling results, we derive design insights and apply them to hardware-specific search space reduction. We show that searching in the reduced search space generates better accuracylatency Pareto frontiers than searching in the original search spaces, customizing architecture search according to the hardware. Moreover, insights derived from measurements lead to notably higher ImageNet top-1 scores on all search spaces investigated.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.