Papers by Gagandeep Singh
The American Surgeon
Hand-assisted techniques facilitated dissemination of the laparoscopic approach in live kidney do... more Hand-assisted techniques facilitated dissemination of the laparoscopic approach in live kidney donors and addressed concerns regarding potential procedural complications. We report our experience with both standard and hand-assisted laparoscopic nephrectomy in routine, complicated, and higher-risk donors. From July 1999 to September 2002, 47 donors underwent standard laparoscopic donor nephrectomy (SLDN; n = 29) or hand-assisted laparoscopic donor nephrectomy (HALDN; n = 18). Donors were “complicated” if they were >60 years of age, obese, refused blood-product transfusion, had multiple renal arteries or veins, or had right nephrectomies. “Higher-risk” donors had two or more risk factors. Results for SLDN and HALDN were compared for the overall groups and for the “complicated” and “higher-risk” groups. No donor required blood transfusion or reoperation. Warm-ischemia times were shorter in left nephrectomies (191 ± 72 seconds vs. 337 ± 95 seconds, P = 0.005), and blood loss was gre...
The American Surgeon
Obstructive jaundice secondary to common bile duct stricture is most commonly attributed to malig... more Obstructive jaundice secondary to common bile duct stricture is most commonly attributed to malignancy. Here we present three unusual cases that mimicked carcinoma in presentation but were histologically diagnosed as benign inflammatory processes during operative care. The first case was attributed to obstruction-induced chronic pancreatitis secondary to Crohn's disease of the head of the pancreas, the second was due to sarcoidosis within periportal and extrahepatic biliary lymph nodes and distal common bile duct, and the third case was due to tuberculosis of biliary lymph nodes. All were successfully managed surgically, but potentially these patients may have been effectively treated pharmacologically, without the need for invasive surgical intervention, if an earlier diagnosis were available to the clinicians. A retrospective and comparative review of the data of each case demonstrated subtle clues such as multiple enlarged biliary lymph node involvement and only moderately el...
Sustainability of Water Resources
Proceedings of the AAAI Conference on Artificial Intelligence
We consider language modelling (LM) as a multi-label structured prediction task by re-framing tra... more We consider language modelling (LM) as a multi-label structured prediction task by re-framing training from solely predicting a single ground-truth word to ranking a set of words which could continue a given context. To avoid annotating top-k ranks, we generate them using pre-trained LMs: GPT-2, BERT, and Born-Again models. This leads to a rank-based form of knowledge distillation (KD). We also develop a method using N-grams to create a non-probabilistic teacher which generates the ranks without the need of a pre-trained LM. We confirm the hypotheses: that we can treat LMing as a ranking task and that we can do so without the use of a pre-trained LM. We show that rank-based KD generally gives a modest improvement to perplexity (PPL) -- though often with statistical significance -- when compared to Kullback–Leibler-based KD. Surprisingly, given the naivety of the method, the N-grams act as competitive teachers and achieve similar performance as using either BERT or a Born-Again model...
CERN European Organization for Nuclear Research - Zenodo, 2004
Apparent molar volumes and viscosities of glycine, DL-a-alanine and L-leucine in water and in 1.0... more Apparent molar volumes and viscosities of glycine, DL-a-alanine and L-leucine in water and in 1.0 m (mol kg-1) aqueous solutions of each of potassium chloride, barium chloride, glucose and sucrose have been determined from density and efflux time measurements, using vibrating-tube digital densitymeter and Ubbelohde viscometer, respectively at 298.15 K. Volume data were utilised to calculate partial molar volumes of transfer at infinite dilution (v~,trl from water to mixed aqueous solutions and B-coefficients were calculated from viscosity data using Jones-Dole equation. Positive v~,tr values have been rationalized using cosphere overlap model in terms of solute-cosolute interactions. Transfer B-coefficients of the amino acids studied are positive in case of glucose and sucrose whereas these are negative in case of electrolytes (KCI, BaCI 2) which have been explained in terms of changes in solvent structure. Contributions of the side chain to ~,tr and to B-coefficient values have also been discussed.
Mechanics And Control Of Large Flexible Structures, 1990
ABSTRACT Two control problems are considered in connection with the minimum-time maneuvers of a f... more ABSTRACT Two control problems are considered in connection with the minimum-time maneuvers of a flexible spacecraft: (1) the time-optimal rest-to-rest slewing problem, which brings the rigid-body mode and a finite number of vibrational modes to rest at the final time; and (2) the time-optimal spin-up problem, which brings a finite number of vibrational modes to rest and leaves the rigid-body mode with a finite velocity at the final time. The control histories are shown to be bang-bang with multiple switches. It is proved that the time-optimal controls for both problems have important time-symmetry properties. A system of nonlinear algebraic equations satisfied by the optimal switching times, the optimal final time, and the optimal costates at midmaneuver is derived. In addition the upper bounds on the spillover are derived for a finite-dimensional evaluation model; and it is observed that, for a scalar control input, the time-optimal control history is independent of the control input location.
Proceedings of the ACM on Programming Languages, 2022
Formal verification of neural networks is critical for their safe adoption in real-world applicat... more Formal verification of neural networks is critical for their safe adoption in real-world applications. However, designing a precise and scalable verifier which can handle different activation functions, realistic network architectures and relevant specifications remains an open and difficult challenge. In this paper, we take a major step forward in addressing this challenge and present a new verification framework, called PRIMA. PRIMA is both (i) general: it handles any non-linear activation function, and (ii) precise: it computes precise convex abstractions involving multiple neurons via novel convex hull approximation algorithms that leverage concepts from computational geometry. The algorithms have polynomial complexity, yield fewer constraints, and minimize precision loss. We evaluate the effectiveness of PRIMA on a variety of challenging tasks from prior work. Our results show that PRIMA is significantly more precise than the state-of-the-art, verifying robustness to input pert...
3 Biotech, 2021
Cyst nematodes of the species Globodera rostochiensis and G. pallida are devastating parasites of... more Cyst nematodes of the species Globodera rostochiensis and G. pallida are devastating parasites of the potato crop. Early detection of cyst nematodes in the field is critical for adopting an appropriate management strategy. A specific and sensitive loop-mediated isothermal amplification (LAMP) assay using four oligonucleotide primers has been developed to amplify the internal transcribed spacer region (ITS) of ribosomal DNA of potato cyst nematode G. rostochiensis. The PCN-LAMP reaction could be completed within 75 min at 68 °C followed by termination at 85 °C for 7 min. The primers exhibited specificity for G. rostochiensis and did not detect any other tested genera of plant parasitic or entomopathogenic nematodes. LAMP reaction was highly sensitive, suitable for crude genomic DNA and could successfully detect G. rostochiensis DNA up to femtogram quantity. This assay is rapid, cost effective and requires minimal instrumentation. It will facilitate the detection of G. rostochiensis at field and point-of-care labs and help in the interception of infested plant material/soil samples at quarantine stations independent of a professional nematologist.
International Journal of Mechanical and Production Engineering Research and Development, 2020
The recent rise in the demand for high precision, close tolerances, and super surface finish qual... more The recent rise in the demand for high precision, close tolerances, and super surface finish quality of components and part assembly modules in the competitive manufacturing industrial environment for the enhanced working life and functional requirements of machines at a global level. A new rotating wheel based magnetorheological process is designed and developed to fine finish the external cylindrical surfaces of soft as well as hard materials. This method is based on the operating concept of the cylindrical grinding system, except that the hard bounded abrasive wheel is replaced with the electromagnet wheel which is capable of generating the regulated magnetic field in the developed process. Due to the generated magnetic field effect, the CIP's particles in the magnetorheological fluid stick-on the wheel and tightly hold the abrasive particles. The fine finishing operation is performed with the help of flexible brushes created by the magnetorheological fluid. A significant range of high precision external cylindrical surface applications can be found in automotive, manufacturing, aerospace, valve & die industries. In the present study, mild steel cylindrical workpiece is used for experimentation purposes. The surface roughness values Ra, Rq, Rz are reduced to 41.17%, 44.62%, 46.45% in 75 minutes of finishing time with rotating wheel based magnetorheological finishing process. The overall results indicate that the new rotating wheel based magnetorheological process is feasible and capable enough to provide fine finishing operation to the external cylindrical surfaces of soft materials as well as hard materials.
Numerical Combustion
Exothermically reactive gas flow in the region between a piston and a strong shock wave, (Mach 3.... more Exothermically reactive gas flow in the region between a piston and a strong shock wave, (Mach 3.0), is modelled numerically using the Random Choice Method to solve the Euler equations. The numerical algorithm consists of the Random Choice scheme, formulated in a Lagrangian configuration, coupled with time-operator splitting to treat combustion chemistry. The combustion reaction is assumed to be of simple irreversible Arrhenius type. Numerical results show that ignition first occurs close to the piston face, and is followed by the formation of a compression pulse that finally contains a shock, an unsteady induction domain and a fast flame prior to transition of this system into a ZND-structured detonation.
Lecture Notes in Computer Science, 2019
Modern scientific workloads have demonstrated the inefficiency of using high precision formats. M... more Modern scientific workloads have demonstrated the inefficiency of using high precision formats. Moving to a lower bit format or even to a different number system can provide tremendous gains in terms of performance and energy efficiency. In this article, we explore the applicability of different number formats and exhaustively search for the appropriate bit width for 3D complex stencil kernels, which are one of the most widely used scientific kernels. Further, we demonstrate the achievable performance of these kernels on state-of-the-art hardware that includes CPU and FPGA, which is the only hardware supporting arbitrary fixed-point precision. Thus, this work fills the gap between current hardware capabilities and future systems for stencil-based scientific applications.
Conference Proceedings. IEEE Canadian Conference on Electrical and Computer Engineering (Cat. No.98TH8341)
ABSTRACT
Proceedings of the 41st ACM SIGPLAN Conference on Programming Language Design and Implementation, 2020
Numerical abstract domains are a key component of modern static analyzers. Despite recent advance... more Numerical abstract domains are a key component of modern static analyzers. Despite recent advances, precise analysis with highly expressive domains remains too costly for many real-world programs. To address this challenge, we introduce a new data-driven method, called Lait, that produces a faster and more scalable numerical analysis without significant loss of precision. Our approach is based on the key insight that sequences of abstract elements produced by the analyzer contain redundancy which can be exploited to increase performance without compromising precision significantly. Concretely, we present an iterative learning algorithm that learns a neural policy that identifies and removes redundant constraints at various points in the sequence. We believe that our method is generic and can be applied to various numerical domains. We instantiate Lait for the widely used Polyhedra and Octagon domains. Our evaluation of Lait on a range of realworld applications with both domains shows that while the approach is designed to be generic, it is orders of magnitude faster on the most costly benchmarks than a state-of-theart numerical library while maintaining close-to-original analysis precision. Further, Lait outperforms hand-crafted heuristics and a domain-specific learning approach in terms of both precision and speed.
Proceedings of IEEE TENCON '98. IEEE Region 10 International Conference on Global Connectivity in Energy, Computer, Communication and Control (Cat. No.98CH36229)
ABSTRACT
Geospatial Technologies for Land and Water Resources Management, 2021
Water Science and Technology Library, 2020
Indian journal of dermatology, 1977
Building neural network classifiers with an ability to distinguish between in and out-of distribu... more Building neural network classifiers with an ability to distinguish between in and out-of distribution inputs is an important step towards faithful deep learning systems. Some of the successful approaches for this, resort to architectural novelties, such as ensembles, with increased complexities in terms of the number of parameters and training procedures. Whereas some other approaches make use of surrogate samples, which are easy to create and work as proxies for actual out-of-distribution (OOD) samples, to train the networks for OOD detection. In this paper, we propose a very simple approach for enhancing the ability of a pretrained network to detect OOD inputs without even altering the original parameter values. We define a probabilistic trust interval for each weight parameter of the network and optimize its size according to the in-distribution (ID) inputs. It allows the network to sample additional weight values along with the original values at the time of inference and use th...
Uploads
Papers by Gagandeep Singh