49 research outputs found
Performance analysis of double pass solar air heater with packed bed porous media in Rajshahi
No full textFunctionalised hexagonal-domain graphene for position-sensitive photodetectors
No full textLetterThis is the author accepted manuscript. The final version is available from IOP Publishing via the DOI in this record.Graphene's unique photoresponse has been largely used in a multitude of optoelectronics applications ranging from broadband photodetectors to wave-guide modulators. In this work we extend the range of applications to position-sensitive photodetectors (PSDs) using FeCl3-intercalated hexagonal domains of graphene grown by atmospheric pressure chemical vapour deposition (APCVD). The FeCl3-based chemical functionalisation of APCVD graphene crystals is affected by the presence of wrinkles and results in a non-uniform doping of the graphene layers. This doping profile creates multiple p–p+ photoactive junctions which show a linear and bipolar photoresponse with respect to the position of a focused light spot, which is ideal for the realization of a PSD. Our study paves the way towards the fabrication of flexible and transparent PSDs that could be embedded in smart textile and wearable electronics.S Russo and M F Craciun acknowledge financial support from EPSRC (Grant no. EP/J000396/1, EP/K017160/1, EP/K010050/1, EPG036101/1, EP/M001024/1, EPM002438/1), from Royal Society international Exchanges Scheme 2016/R1, from European Commission (FP7-ICT-2013-613024-GRASP) and from the Leverhulme Trust (grant title 'Quantum Drums' and 'Room temperature quantum electronics'). I Amit received funding from the People Programme (Marie Curie Actions) of the European Union's Eighth Framework Programme Horizon 2020 under REA grant agreement number 701704
Optical properties of Type-I GaAsP/AlGaAs nano-heterostructure under external uniaxial strain
No full textDesign modification of airfoil by integrating sinusoidal leading edge and dimpled surface
No full textCompetition in mortgage markets: the effect of lender type on loan characteristics
Get PDFThis article examines how competition among lenders affects mortgage loan characteristics. The author finds that, on average, banks issue safer mortgages than independent mortgage banks. Further, mortgages from banks with a branch in the local market where the property is tend to be safer than mortgages from banks without a local branch. Changes in market shares among lender types (local bank, nonlocal bank, or independent mortgage bank) that lead to higher loan risk also are associated with better borrower quality. Increasing the local market share of a lender type raises loan risk and borrower quality at that lender type.
Competition in mortgage markets: the effect of lender type on loan characteristics
Get PDFThis article examines how competition among lenders affects mortgage loan characteristics. The author finds that, on average, banks issue safer mortgages than independent mortgage banks. Further, mortgages from banks with a branch in the local market where the property is tend to be safer than mortgages from banks without a local branch. Changes in market shares among lender types (local bank, nonlocal bank, or independent mortgage bank) that lead to higher loan risk also are associated with better borrower quality. Increasing the local market share of a lender type raises loan risk and borrower quality at that lender type.Mortgage loans ; Mortgages ; Home Mortgage Disclosure Act
Knowledge-Infused Learning: A Sweet Spot in Neuro-Symbolic AI
No full textDeep learning has revolutionized the artificial intelligence (AI) landscape by enhancing machine capabilities to understand data-dependant relationships. On the other hand, knowledge may not directly correlate or depend on the data but represents facts that are true. Combining knowledge with the data-driven deep learning techniques improves upon what can be learned from data alone, resulting in improved performance with reduced training, user-level explainability, modeling uncertainty in deep learning, achieving context-sensitivity, and better control over the behavior of AI systems such as to assure the safety or avoid toxic behavior. We refer to the approach of combining various types of explicit knowledge as knowledge-infused learning (KiL). Knowledge infusion brings symbolic AI into data-driven AI, giving us a class of neuro-symbolic AI methods. The work on KiL has already developed a suite of context-adaptive algorithms that infuses various knowledge into deep learning methods in various ways, broadly categorized as a shallow infusion, semi-deep infusion, and deep infusion. This special issue allows interdisciplinary researchers and practitioners from diverse fields such as natural language processing, recommender systems, and computer vision to contribute their research on the infusion of external and expert-curated knowledge in data-driven learning methodologies for consistency and robustness in outcomes.This work was supported by the National Science Foundation (NSF) Award #2133842 “EAGER: Advancing Neuro-Symbolic AI with Deep Knowledge-infused Learning.” Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF, Samsung Research America, and Amazon.https://ieeexplore.ieee.org/document/984141
Intersection modifications using mini-/modular-roundabout methods
No full textFinal report; "November 2021."; Additional project information provided in email with PDFs: SJN 135995; Author Md Amdad Hossen's name is misspelled on the cover page, technical report documentation page, and title page of the final report as well as on the fact sheet; Includes bibliographical references (pages 22-26 of Final report)Final report (viii, 188 pages) -- [Fact sheet] (2 unnumbered pages)Conversion of traditional intersections (stop controlled and signalized) to modern RABs has been a growing practice in many countries around the world including the U.S. - largely due to the benefits of reduction in crash frequency and severity, capacity improvement, and operational improvement. However, construction of traditional RABs is costly and requires additional right-of-way (ROW) which can deter roundabout installation on local transportation systems that have budgetary and/or available ROW restrictions. The main objective of this project was to develop guidelines for ORIL on the installation and performance of mini-/modular-RABs considering characteristics of Ohio's local transportation system. Based on published guidelines and from existing pilot implementations (both international and within U.S.), current design practices considering traffic condition and roadway conditions were identified. Based on survey findings, there was a reasonably high level of familiarity with mini-RABs among respondents. Most agencies consider reduction of crashes/severity and improved traffic operations in installation of mini-RABs. Major concern with mini-RABs is drivers neglecting the central island and driving straight through thus causing the mini-/modular-RAB to lose its integrity. Agencies typically place mini-/modular-RABs on two-lane highways with low traffic volumes (<15,000 vpd); and/or peak-hour volumes of 1,600 to 1,800 vehicles. Based on driving simulator experiments, there are no differences in critical gap as driver's maneuvered through mini-RAB of different ICDs. Operations-wise, Mini-RABs with larger ICDs performed better than those with smaller ICD
REASONS: A benchmark for REtrieval and Automated citationS Of scieNtific Sentences using Public and Proprietary LLMs
No full textAutomatic citation generation for sentences in a document or report is paramount for intelligence analysts, cybersecurity, news agencies, and education personnel. In this research, we investigate whether large language models (LLMs) are capable of generating references based on two forms of sentence queries: (a) Direct Queries, LLMs are asked to provide author names of the given research article, and (b) Indirect Queries, LLMs are asked to provide the title of a mentioned article when given a sentence from a different article. To demonstrate where LLM stands in this task, we introduce a large dataset called REASONS comprising abstracts of the 12 most popular domains of scientific research on arXiv. From around 20K research articles, we make the following deductions on public and proprietary LLMs: (a) State-of-the-art, often called anthropomorphic GPT-4 and GPT-3.5, suffers from high pass percentage (PP) to minimize the hallucination rate (HR). When tested with Perplexity.ai (7B), they unexpectedly made more errors; (b) Augmenting relevant metadata lowered the PP and gave the lowest HR; (c) Advance retrieval-augmented generation (RAG) using Mistral demonstrates consistent and robust citation support on indirect queries and matched performance to GPT-3.5 and GPT-4. The HR across all domains and models decreased by an average of 41.93%, and the PP was reduced to 0% in most cases. In terms of generation quality, the average F1 Score and BLEU were 68.09% and 57.51%, respectively; (d) Testing with adversarial samples showed that LLMs, including the Advance RAG Mistral, struggle to understand context, but the extent of this issue was small in Mistral and GPT-4-Preview. Our study contributes valuable insights into the reliability of RAG for automated citation generation tasks.http://arxiv.org/abs/2405.0222
