Keyword assignment is an important step towards semantic enablement of the web. In this paper we describe a taxonomy called Agrotags which is designed for tagging agriculture documents. Agrotags is a subset of Agrovoc and is much smaller: about 2100 as against 40,000. Agrotags is manually created by carefully examining each of the Agrovoc terms for their utility in tagging. This selected subset is further refined and validated by looking at the manually assigned keywords from Agris databases. Further extending the usage of Agrotags emerges the concept of Agrotagger which is a system for automatically generating keywords for agricultural documents. Agrotagger has been built by moving the learning (what keyword to assign) from the example (document) level to the model level. Agrotagger being a pluggable module can act as an add-on to any repository.
No abstract
<p>Power and signaling infrastructure comprise of electrical installations, electronic systems and cables. A damage to these cables will be catastrophic or at the least significant downtime to the system. Hence, monitoring these cables in real-time not only improves the efficiency of the system but can also avoid fatal accidents. In this work, we develop a non-invasive composite diagnostic framework to identify cable damages such as insulation cuts. The framework can detect, classify, and locate faults. While the solution is general enough, we consider two use cases: (a) railway cables used in signalling applications and(b) symmetrical four-core cables used in residential buildings. In order to characterize the faults, we use three non-invasive sensors: (a) SNR sensor, (b) S-parameter, and (c) Correlation peak sensor. We use a single programmable hardware to implement each of these sensors. These sensors monitor a parameter change on the cable in real-time. The experimental insights gained are used to construct an a priori Bayesian network depicting the non-deterministic relationship between an effect and its causes. This uncertainty is due to inhibitors such as sensor calibration issues and coupling mismatch between sensor transceivers suitably handled through a Noisy-OR function. The results of Bayesian inference with belief propagation provides up to 97% match with the ground truth state. Regarding the cable fault, our experimental results show a best-case detection and localization accuracy of 98% & 97.2% respectively.</p>
Federated learning (FL) has evolved as a prominent method for edge devices to cooperatively create a unified prediction model while securing their sensitive training data local to the device. Despite the existence of numerous research frameworks for simulating FL algorithms, they do not facilitate comprehensive deployment for automatic speech recognition tasks on heterogeneous edge devices. This is where Ed-Fed, a comprehensive and generic FL framework, comes in as a foundation for future practical FL system research. We also propose a novel resource-aware client selection algorithm to optimise the waiting time in the FL settings. We show that our approach can handle the straggler devices and dynamically set the training time for the selected devices in a round. Our evaluation has shown that the proposed approach significantly optimises waiting time in FL compared to conventional random client selection methods.
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.