3,556 research outputs found
IISc-DIO
No full textData and codes used in 'Pramod, R. T., & Arun, S. P. (2016). Do computational models differ systematically from human object perception?. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 1601-1609)
IISc-DIO
No full textData and codes used in 'Pramod, R. T., & Arun, S. P. (2016). Do computational models differ systematically from human object perception?. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 1601-1609)
Excited-state intramolecular proton transfer of 2-acetylindan-1,3-dione studied by ultrafast absorption and fluorescence spectroscopy
No full textWe employ transient absorption from the deep-UV to the visible region and fluorescence upconversion to investigate the photoinduced excited-state intramolecular proton-transfer dynamics in a biologically relevant drug molecule, 2-acetylindan- 1,3-dione. The molecule is a ß-diketone which in the electronic ground state exists as exocyclic enol with an intramolecular H-bond. Upon electronic excitation at 300 nm, the first excited state of the exocyclic enol is initially populated, followed by ultrafast proton transfer (±160 fs) to form the vibrationally hot endocyclic enol. Subsequently, solvent-induced vibrational relaxation takes place (±10 ps) followed by decay (±390 ps) to the corresponding ground state. © 2015 Author(s)1561sciescopu
Compositionality of Object Representations in Brains and Machines
No full textCompositionality in object vision can be defined as the principles governing the
relationship between whole objects and their constituent attributes. It is known that object
information falling on the retina is processed in a hierarchy of cortical regions starting from
simple edge-detectors in the primary visual cortex to complex shape representations in the
higher visual cortex, yet we still do not understand how whole objects are represented in terms
of their attributes. With recent advances in computer vision, we have, for the first time in
history, a very good machine vision system in the form of convolutional neural networks. How
do these systems compare with human vision? We argue that understanding vision in the brain
and making machines see the way we do form two sides of the same coin – understanding one
will give us insights into the other. With this in mind, the goal of my thesis is twofold – to
study compositionality in object representations in the brain; and to compare compositionality
in brains and machines with the goal of improving machine vision.
I will present results from a series of studies where we investigate object representations
in brains and machines. In the first set of studies, we investigated whether whole object
responses in perception and in single neurons could be understood in terms of their parts. The
main findings are: (1) Object attributes combine linearly in visual search (Pramod & Arun,
2016); (2) Although symmetry is a salient holistic property, responses to symmetric objects are
also explained as a sum of their parts as were asymmetric objects (Pramod & Arun, 2018).
Taken together these findings confirm the compositionality of object representations in
perception and in high-level visual cortex.
In the second set of studies, we compared the compositionality of object representations
in brains and machines. The main findings are: (1) Object representations in virtually all
computer vision models (including deep neural networks) deviate systematically from human
perception (Pramod & Arun, 2016); (2) Symmetric objects are more salient in perception than
in deep neural networks, and fixing this bias leads to significant improvements in object
detection performance; and finally, (3) we show that under-sampling of the periphery in the
biological retina is computationally optimal for object recognition in natural scenes, pointing
to dissociable roles for object and context. Taken together, these findings show that machine
vision can be understood and improved by studying biological vision
Peperomia mangalbaria (Piperaceae), a new species from Sikkim Himalaya, India
No full textRai, Pramod, Mathieu, Guido (2023): Peperomia mangalbaria (Piperaceae), a new species from Sikkim Himalaya, India. Phytotaxa 609 (2): 138-144, DOI: 10.11646/phytotaxa.609.2.6, URL: http://dx.doi.org/10.11646/phytotaxa.609.2.
Supersymmetric many-body systems from partial symmetries — integrability, localization and scrambling
No full textPartial symmetries are described by generalized group structures known as symmetric inverse semigroups. We use the algebras arising from these structures to realize supersymmetry in (0+1) dimensions and to build many-body quantum systems on a chain. This construction consists in associating appropriate supercharges to chain sites, in analogy to what is done in spin chains. For simple enough choices of supercharges, we show that the resulting states have a finite non-zero Witten index, which is invariant under perturbations, therefore defining supersymmetric phases of matter protected by the index. The Hamiltonians we obtain are integrable and display a spectrum containing both product and entangled states. By introducing disorder and studying the out-of-time-ordered correlators (OTOC), we find that these systems are in the many-body localized phase and do not thermalize. Finally., we reformulate a theorem relating the growth of the second Rényi entropy to the OTOC on a thermal state in terms of partial symmetries. © 2017, The Author(s)2211Nsciescopu
Neural geolocation prediction in Twitter
No full textInferring the location of a user has been a valuable step for many applications that leverage social media, such as marketing, security monitoring and recommendation systems. Motivated by the recent success of Deep Learning techniques for many tasks such as computer vision, speech recognition, and natural language processing, we study the application of neural models to the problem of geolocation prediction and experiment with multiple techniques to analyze neural networks for geolocation inference based solely on text. Experimental results on the dataset suggest that choosing appropriate network architecture can all increase performance on this task and demonstrate a promising extension of neural network based models for geolocation prediction. Our systematic extensive study of four supervised and three unsupervised tweet representations reveal that Convolutional Neural Networks (CNNs) and fastText best encode the the textual and geoloca- tional properties of tweets respectively. fastText emerges as the best model for low resource settings, providing very little degradation with reduction in embedding size.Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2019-05-01The student, Pramod Srinivasan, accepted the attached license on 2017-04-25 at 12:15.The student, Pramod Srinivasan, submitted this Thesis for approval on 2017-04-25 at 12:51.This Thesis was approved for publication on 2017-04-25 at 18:42.DSpace SAF Submission Ingestion Package generated from Vireo submission #11043 on 2017-08-10 at 14:32:36Made available in DSpace on 2017-08-10T19:52:23Z (GMT). No. of bitstreams: 2
SRINIVASAN-THESIS-2017.pdf: 1215687 bytes, checksum: 96dbc159bb19eab4d69b3df1dfcffd17 (MD5)
LICENSE.txt: 4214 bytes, checksum: 6d429007259258d1f9571b8e0eac0cf7 (MD5)
Previous issue date: 2017-04-25Embargo set by: Colleen Fallaw for item 102685
Lift date: 2019-08-10T21:25:30Z
Reason: Author requested closed access (OA after 2yrs) in Vireo ETD systemLimited Restriction Lifted for Item 102685 on 2019-08-11T09:15:17Z
Structural and electronic properties of amorphous Ti-Ni alloy thin films prepared by ion beam sputtering
No full text
Similitude of membrane helical coil with membrane serpentine tube for characteristics of high-pressure syngas: A review
No full text- …
