163 research outputs found
Factors and Activities Considered by First Generation Agripreneurs for Agri-Business Sustainable Development: A Study of Haryana, India
No full textThe main objective of the present study is to identify the factors and activities considered by first generation agripreneuers in managing agribusiness which plays a vital role in the success and sustainable development of any agribusiness. To accomplish this objective, a factor analysis method was adopted to gather and understand the findings. The data has been gathered from four geographical zones (East zone, West zone, North zone, South zone) of Haryana state comprising 22 districts through a purposive sampling strategy. The major findings of the study highlight that out of seven factors, effective leadership was considered the most important factor with the highest Cronbach value 0.812. It plays significant role in the success of any enterprise by ensuring supportive environment for the workers in agro-industries. Strategic planning was also important because becoming a successful agripreneur requires planning before converting all business activities into action.In addition, all the other factors such as scanning business opportunities, organizing and business activities, prior analysis, and credit facilities all play a vital role in the success of agro-industries
Face identification and clustering
No full textIn this thesis, we study two problems based on clustering algorithms. In the first problem, we study the role of visual attributes using an agglomerative clustering algorithm to whittle down the search area where the number of classes is high to improve the performance of clustering. We observe that as we add more attributes, the clustering performance increases overall. In the second problem, we study the role of clustering in aggregating templates in a 1:N open set protocol using multi-shot video as a probe. We observe that by increasing the number of clusters, the performance increases with respect to the baseline and reaches a peak, after which increasing the number of clusters causes the performance to degrade. Experiments are conducted using recently introduced unconstrained IARPA Janus IJB-A, CS2, and CS3 face recognition datasets.M.S.Includes bibliographical referencesby Atul Dhingr
Self-assembled strained nanostructures for light emission grown using molecular beam epitaxy
Get PDFIII-V nanostructures are widely researched for applications in dislocation-resistant light emitters for photonic integrated circuits, quantum computing and single photon emitters. The 0D nanostructures include quantum dots (QDs), dot in a well (DWELLs), sub-monolayer QDs and droplet epitaxy QDs, while 1D elongated structures include quantum dashes and nanowires (NWs). The optical properties of nanostructures can be controlled through size, composition, strain and band-offsets during epitaxial growth and can be tailored precisely to emit light with photon energies suited to the application, spanning 0.2-2.0 eV. This thesis explores two novel QD based light emitters in the visible and near-infrared wavelength regime. In the first part of the thesis, we demonstrate the growth and characterization of tensile strained Ge QDs and Ge NWs phase segregated in the III-V matrix via Volmer-Weber growth mode emitting at 1200 nm. The second part of the thesis demonstrates the dislocation tolerance of compressively strained InP QDs grown on lattice-matched GaAs and lattice-mismatched Si substrate via Stranski-Krastanov growth mode emitting at 713 nm.
The first part of the thesis explores the growth of tensile strained Ge QDs and NWs phase segregated in the III-V matrix. Epitaxial growth of phase segregated Ge nanostructures embedded within III-V compound semiconductors is a promising way to achieve a high biaxial tensile strain along with precise control of nanostructure density, size and morphology. Here we demonstrate growth of phase-segregated Ge quantum dots (QDs) and compare them to our previously reported Ge nanowires (NWs); both are strained to an In0.52Al0.48As matrix with a high biaxial tensile strain of 3.6%. Despite the similar growth conditions, there exist pronounced differences in the lateral size and planar density of Ge QDs and Ge NWs, with Ge QDs showing significantly larger size, lower density and structural anisotropy along the in-plane [1-10] direction. In addition to the difference in morphology, Ge QDs are shown to be more prone to plastic relaxation by formation of dislocations and stacking faults, which we attribute to their larger in-plane size. Finally, tensile Ge QDs are shown to exhibit strong room-temperature photoluminescence at 1176 nm, which is blueshifted from the case of Ge NWs.
In the second part of the thesis, we demonstrate epitaxial InP QDs on GaAs on Si virtual substrates with room-temperature photoluminescence (PL) intensity nearly identical to those grown on GaAs substrates. The similarity in PL characteristics is remarkable considering that the active region on the GaAs/Si virtual substrate has a threading dislocation density (TDD) of ~3×10^7 cm-2, as compared to the bulk GaAs substrate with TDD 50× improvement in the luminescence intensity of InP QDs annealed at ~700⁰C for 100 minutes without observable structural degradation or blue-shift in the PL spectrum.Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2021-05-01The student, Pankul Dhingra, accepted the attached license on 2019-04-25 at 12:06.The student, Pankul Dhingra, submitted this Thesis for approval on 2019-04-25 at 12:16.This Thesis was approved for publication on 2019-04-25 at 14:10.DSpace SAF Submission Ingestion Package generated from Vireo submission #13914 on 2019-08-22 at 16:23:56Made available in DSpace on 2019-08-23T20:48:26Z (GMT). No. of bitstreams: 2
DHINGRA-THESIS-2019.pdf: 2735717 bytes, checksum: 55584f4a818d3f00a92b3ad38753e24d (MD5)
LICENSE.txt: 4211 bytes, checksum: 108fd1426b2a5d615ea1ebad7d58e69f (MD5)
Previous issue date: 2019-04-25Embargo set by: Seth Robbins for item 112387
Lift date: 2021-08-23T20:48:32Z
Reason: Author requested closed access (OA after 2yrs) in Vireo ETD systemLimited Restriction Lifted for Item 112387 on 2021-08-24T09:15:38Z
Model Complexity-Accuracy Trade-off for a Convolutional Neural Network
Get PDFConvolutional Neural Networks(CNN) has had a great success in the recent
past, because of the advent of faster GPUs and memory access. CNNs are really
powerful as they learn the features from data in layers such that they exhibit
the structure of the V-1 features of the human brain. A huge bottleneck, in
this case, is that CNNs are very large and have a very high memory footprint,
and hence they cannot be employed on devices with limited storage such as
mobile phone, IoT etc. In this work, we study the model complexity versus
accuracy trade-off on MNSIT dataset, and give a concrete framework for handling
such a problem, given the worst case accuracy that a system can tolerate. In
our work, we reduce the model complexity by 236 times, and memory footprint by
19.5 times compared to the base model while achieving worst case accuracy
threshold
Face Identification and Clustering
Get PDFIn this thesis, we study two problems based on clustering algorithms. In the
first problem, we study the role of visual attributes using an agglomerative
clustering algorithm to whittle down the search area where the number of
classes is high to improve the performance of clustering. We observe that as we
add more attributes, the clustering performance increases overall. In the
second problem, we study the role of clustering in aggregating templates in a
1:N open set protocol using multi-shot video as a probe. We observe that by
increasing the number of clusters, the performance increases with respect to
the baseline and reaches a peak, after which increasing the number of clusters
causes the performance to degrade. Experiments are conducted using recently
introduced unconstrained IARPA Janus IJB-A, CS2, and CS3 face recognition
datasets
Indian Electoral Roll PDF Corpus
No full textThe data repository contains PDFs of the Indian electoral rolls. The data were primarily collected in 2017. The scripts used to scrape the data and the metadata about the data are on Github. The parsed data is available here (under the same access conditions as this data). The scripts used to parse the data are posted on Github
Instate: Predict the state from last name
No full textThe data and model artifacts posted to the repository form the basis of the Python package Instate. The paper associated with the Python package is posted to the GitHub repository.
The data in this repository contains the elector name and state. The data uses the Parsed Indian Electoral Rolls Corpus
which in turn uses the Indian Electoral Roll PDF Corpus
Scaling ML Products At Startups: A Practitioner's Guide
Get PDFHow do you scale a machine learning product at a startup? In particular, how
do you serve a greater volume, velocity, and variety of queries
cost-effectively? We break down costs into variable costs-the cost of serving
the model and performant-and fixed costs-the cost of developing and training
new models. We propose a framework for conceptualizing these costs, breaking
them into finer categories, and limn ways to reduce costs. Lastly, since in our
experience, the most expensive fixed cost of a machine learning system is the
cost of identifying the root causes of failures and driving continuous
improvement, we present a way to conceptualize the issues and share our
methodology for the same
- …
