2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020
Landmark detection algorithms trained on high resolution images perform poorly on datasets contai... more Landmark detection algorithms trained on high resolution images perform poorly on datasets containing low resolution images. This degrades the performance of facial verification, recognition and modeling that rely on accurate detection of landmarks. To the best of our knowledge, there is no dataset consisting of low resolution face images along with their annotated landmarks, making supervised training infeasible. In this paper, we present a semi-supervised approach to predict landmarks on low resolution images by learning them from labeled high resolution images. The objective of this work is to show that predicting landmarks directly on low resolution images is more effective than the current practice of aligning images after rescaling or super-resolution. In a two-step process, the proposed approach first learns to generate low resolution images by modeling the distribution of target low resolution images. In the second stage, the model learns to predict landmarks for target low ...
Information and Communication Technology for Sustainable Development
Visual analytics uses interactive visualizations in order to incorporate user’s knowledge and cog... more Visual analytics uses interactive visualizations in order to incorporate user’s knowledge and cognitive capability into data analytics processes. The progressive visual analytic paradigm simplifies the analytic process when it comes to large datasets. It uses the interactive sequential pattern mining algorithm which reports patterns as it finds them. But, the sequential pattern mining algorithms like SPAM, SPADE and PrefixSpan are suited for a single-node environment only. It is also constrained by the available size of memory and computational power while handling a very large quantity of data. So to overcome these challenges, the proposed MapReduce frequent pattern mining (MR-FPM) algorithm distributes data across various nodes in the Hadoop cluster, finds the candidate itemsets and counts their support using the MapReduce paradigm. The patterns with supportless than the user-defined minsup are discarded. Experimental results show that MR-FPM continuously outperforms SPAM when the minsup is decreased.
The aggregates of a protein called, 'Aβ' found in brains of Alzheimer's patients a... more The aggregates of a protein called, 'Aβ' found in brains of Alzheimer's patients are strongly believed to be the cause for neuronal death and cognitive decline. Among the different forms of Aβ aggregates, smaller aggregates called 'soluble oligomers' are increasingly believed to be ...
Congestion management is one of the technical challenges in power system deregulation. This paper... more Congestion management is one of the technical challenges in power system deregulation. This paper presents single objective and multi-objective optimization approaches for optimal choice, location and size of Static Var Compensators (SVC) and Thyristor Controlled Series Capacitors (TCSC) in deregulated power system to improve branch loading (minimize congestion), improve voltage stability and reduce line losses. Though FACTS controllers offer many advantages, their installation cost is very high. Hence Independent System Operator (ISO) has to locate them optimally to satisfy a desired objective. Genetic Algorithms (GA) are best suitable for solution of combinatorial optimization and multi-objective optimization problems. This paper presents optimal location of FACTS controllers considering branch loading (BL), voltage stability (VS) and loss minimization (LM) as objectives at once using GA. It is observed that the locations that are most favorable with respect to one objective are n...
The study of the effect of priced information on basic al-gorithmic problems was initiated by the... more The study of the effect of priced information on basic al-gorithmic problems was initiated by the paper of Charikar et al. [5]. In this paper, we continue the study of sorting and selection in the priced comparison model, ie, when each comparison has an associated cost, and answer ...
Abstract. We consider the design of resilient networks that are fault tolerant against link failu... more Abstract. We consider the design of resilient networks that are fault tolerant against link failures. Resilience against link failures can be built into the network by providing backup paths, which are used in the eventuality of an edge failure occurring on a primary path in ...
We study the multicommodity rent-or-buy problem, a type of network design problem with economies ... more We study the multicommodity rent-or-buy problem, a type of network design problem with economies of scale. In this problem, capacity on an edge can be rented, with cost incurred on a per-unit of capacity basis, or bought, which allows unlimited use after payment of a large fixed ...
Abstract. We present approximation algorithms for the unsplittable flow problem (UFP) in undirect... more Abstract. We present approximation algorithms for the unsplittable flow problem (UFP) in undirected graphs. As is standard in this line of research, we assume that the maximum demand is at most the minimum capacity. We focus on the non-uniform capacity case in which the edge ...
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020
Landmark detection algorithms trained on high resolution images perform poorly on datasets contai... more Landmark detection algorithms trained on high resolution images perform poorly on datasets containing low resolution images. This degrades the performance of facial verification, recognition and modeling that rely on accurate detection of landmarks. To the best of our knowledge, there is no dataset consisting of low resolution face images along with their annotated landmarks, making supervised training infeasible. In this paper, we present a semi-supervised approach to predict landmarks on low resolution images by learning them from labeled high resolution images. The objective of this work is to show that predicting landmarks directly on low resolution images is more effective than the current practice of aligning images after rescaling or super-resolution. In a two-step process, the proposed approach first learns to generate low resolution images by modeling the distribution of target low resolution images. In the second stage, the model learns to predict landmarks for target low ...
Information and Communication Technology for Sustainable Development
Visual analytics uses interactive visualizations in order to incorporate user’s knowledge and cog... more Visual analytics uses interactive visualizations in order to incorporate user’s knowledge and cognitive capability into data analytics processes. The progressive visual analytic paradigm simplifies the analytic process when it comes to large datasets. It uses the interactive sequential pattern mining algorithm which reports patterns as it finds them. But, the sequential pattern mining algorithms like SPAM, SPADE and PrefixSpan are suited for a single-node environment only. It is also constrained by the available size of memory and computational power while handling a very large quantity of data. So to overcome these challenges, the proposed MapReduce frequent pattern mining (MR-FPM) algorithm distributes data across various nodes in the Hadoop cluster, finds the candidate itemsets and counts their support using the MapReduce paradigm. The patterns with supportless than the user-defined minsup are discarded. Experimental results show that MR-FPM continuously outperforms SPAM when the minsup is decreased.
The aggregates of a protein called, 'Aβ' found in brains of Alzheimer's patients a... more The aggregates of a protein called, 'Aβ' found in brains of Alzheimer's patients are strongly believed to be the cause for neuronal death and cognitive decline. Among the different forms of Aβ aggregates, smaller aggregates called 'soluble oligomers' are increasingly believed to be ...
Congestion management is one of the technical challenges in power system deregulation. This paper... more Congestion management is one of the technical challenges in power system deregulation. This paper presents single objective and multi-objective optimization approaches for optimal choice, location and size of Static Var Compensators (SVC) and Thyristor Controlled Series Capacitors (TCSC) in deregulated power system to improve branch loading (minimize congestion), improve voltage stability and reduce line losses. Though FACTS controllers offer many advantages, their installation cost is very high. Hence Independent System Operator (ISO) has to locate them optimally to satisfy a desired objective. Genetic Algorithms (GA) are best suitable for solution of combinatorial optimization and multi-objective optimization problems. This paper presents optimal location of FACTS controllers considering branch loading (BL), voltage stability (VS) and loss minimization (LM) as objectives at once using GA. It is observed that the locations that are most favorable with respect to one objective are n...
The study of the effect of priced information on basic al-gorithmic problems was initiated by the... more The study of the effect of priced information on basic al-gorithmic problems was initiated by the paper of Charikar et al. [5]. In this paper, we continue the study of sorting and selection in the priced comparison model, ie, when each comparison has an associated cost, and answer ...
Abstract. We consider the design of resilient networks that are fault tolerant against link failu... more Abstract. We consider the design of resilient networks that are fault tolerant against link failures. Resilience against link failures can be built into the network by providing backup paths, which are used in the eventuality of an edge failure occurring on a primary path in ...
We study the multicommodity rent-or-buy problem, a type of network design problem with economies ... more We study the multicommodity rent-or-buy problem, a type of network design problem with economies of scale. In this problem, capacity on an edge can be rented, with cost incurred on a per-unit of capacity basis, or bought, which allows unlimited use after payment of a large fixed ...
Abstract. We present approximation algorithms for the unsplittable flow problem (UFP) in undirect... more Abstract. We present approximation algorithms for the unsplittable flow problem (UFP) in undirected graphs. As is standard in this line of research, we assume that the maximum demand is at most the minimum capacity. We focus on the non-uniform capacity case in which the edge ...
Uploads
Papers by AMIT KUMAR