Year-2018
http://repository.iiitd.edu.in/xmlui/handle/123456789/612
2024-03-28T18:42:42ZPickMe : task allocation in mobile crowdsensing
http://repository.iiitd.edu.in/xmlui/handle/123456789/721
PickMe : task allocation in mobile crowdsensing
Bajaj, Garvita; Singh, Pushpendra (Advisor)
I am eternally grateful to a lot of people who, in no particular order, have helped me get through the tough times during the course of this dissertation. I have no words to thank my lovely parents who have always taught me to be independent and live a happy life. My father, Mr H
C Bajaj, has forever believed in me and continues to encourage me to open my wings to fly.
His trust for me overshadows his overprotecting nature, which is the reason I had the courage to pursue higher studies. While my father’s belief gave me wings to fly, my mother, Mrs Anju Bajaj, always motivated me while keeping me grounded. She is my mentor, my guardian angel, my friend, my support-system who has taught me to deal with reality. My parents have worked together to shape me as a philomath while teaching me the importance of being available for the family. I would also like to express my gratitude to my forever angry (read “hulk”) brother Kush and my soul sister Asha, who deserve a special mention for always reminding me that it’s okay to fail sometimes, but it’s equally important to have fun and laugh at those failures.
I have no words to thank my advisor, Dr. Pushpendra Singh, who invested his time in training me as a researcher and believed in me more than myself. As a guide, his knowledge and his ability to deal patiently with all my questions and doubts (no matter how silly), has not only eased my learning but also taught me that it’s important to ask questions and define priorities.
On the personal front, he has always encouraged all members of CoDes lab, irrespective of their gender, to stay strong and not give up anything because of any sort of societal pressures. He has truly led us all by example. His thought process and societal outlook is something that I admire, and I hope to follow in his footsteps.
I would also like to thank MiMove team at Inria, Paris, especially, my mentors, Dr. Animesh
Pathak and Dr. Rachit Agarwal. Animesh left no stone unturned to teach me the importance of
“fail-fast” approach. He showed me the limitless possibilities that technology brings forth. His passion to quickly define problems and find solutions is something that inspires almost every researcher that he comes across. I am equally indebted to the other members of MiMove team and friends from Paris who truly helped me enjoy the European experience while learning new things. And, thank you, Ashish! You have no idea how good a friend you are. If not for ‘Apna
Punjab’, my brain and heart would have never got the Indian food they needed to keep going there!
My friends and colleagues have been no less than family during this long journey. Starting with my M.Tech. days, everyone here has been a mentor for me in one way or the other. I learned a lot of traits from people around me: professionalism from Parikshit and Deepika, discipline from
Haroon, dedication from Alvika and Samy, commitment from Milan and Dheryta, hard-work from Samy, smart-work from Siddhartha and Nipun, and the way of living from Anupriya and Jasmeet. I am fortunate to have these people around me. They made this the most enjoyable and memorable time of my life.
Lastly, to the latest addition in my life - Ravi Maggon, thank you for being there! Thank you for believing in me, and thank you for helping me stay calm during the toughest phase of this journey. I am glad you are here to stay forever now.
Because of these special people, it took years to write this thesis. Without them, it would have taken decades...if not forever.
2018-01-01T00:00:00ZDetecting anomalous energy consumption in buildings using smart meter data
http://repository.iiitd.edu.in/xmlui/handle/123456789/706
Detecting anomalous energy consumption in buildings using smart meter data
Lone, Haroon Rashid; Singh, Pushpendra (Advisor)
Buildings consume 50% of the total available electrical energy. Studies show that up to 20% of the energy gets waste due to several reasons such as appliance misconfigurations, faults, and abnormal user behavior. The instances of abnormal energy consumption by various electrical appliances are known as anomalies, and it is important to detect such anomalies in a timely manner.
Several works use intrusive appliance-level monitors, i.e., submetering, for detecting
anomalous behavior of appliances. These approaches detect anomalies, but the intrusive monitoring approach is not scalable as it requires to have a separate monitor for each appliance in a home. A more practical approach is to use the household's aggregate consumption data from a single smart meter. Existing methods using smart meter data suffer from two limitations: (i) they result in a high number of false alarms which makes them unreliable, (ii) they only detect anomalies and do not identify the anomaly causing appliance, which makes it difficult for building administrator to take prompt action.
Using smart meter data, we handled the problem of anomaly detection both at coarse and _ne-granular levels. At coarse-level, we proposed a density-based anomaly detection approach namely, Monitor, which takes the power readings of several days as input and outputs anomaly score for each day consumption. At the granular level, we proposed an anomaly detection approach namely, RIMOR, and also evaluated the efficacy of Non-Intrusive Load Monitoring (NILM) for identifying anomalous appliance. Results show that proposed approaches reduce the false alarm rate considerably and also identify the anomalous appliances.
We did all our experiments on publicly available datasets namely Dataport, REFIT, AMPds, and ECO. These datasets have both aggregate and appliance-level energy consumption data, but none of them provides information about anomalous instances. We created and publicly released detailed anomaly annotations for REFIT at appliance-level and the remaining three datasets at the aggregate-level. Furthermore, we collected a context-rich four-year energy dataset from our campus and made it publicly available so that it can be leveraged to propose newer anomaly detection approaches with higher accuracy and take the research in a new direction.
2018-12-01T00:00:00ZForensics enabled secure mobile computing system for enterprises
http://repository.iiitd.edu.in/xmlui/handle/123456789/704
Forensics enabled secure mobile computing system for enterprises
Govindaraj, Jayaprakash; Gupta, Gaurav (Advisor); Chang, Donghoon (Advisor)
The enterprises are facing constant security threats with the emergence of new mobile computing devices like smartphones, smartwatches, and wearables. The existing digital forensic enabled security solutions are not able to match the pace at which these technological advances are evolving. In case of a security incident, the enterprises fail to perform a subsequent digital forensic investigation since their security systems are not designed with required Digital Forensic Readiness capability. Currently, there are no well defined Digital Forensic Readiness frameworks for mobile computing devices. The are some existing frameworks that provide partial support. However, they do not have a provision to learn from the past security violation occurrences. There is a need for an automated forensically ready and secure solution, which could improve efficiency and productivity, while continuously learning and adapting to new and unforeseen challenges. The current thesis is devoted to the design of forensics enabled secure mobile computing systems for the enterprises. The author has focused on developing a ‘digital forensic readiness and secure’ system, which targets smartphones, smartwatches, and wearables operating in an enterprise environment; while incorporating machine learning capabilities to make it a learning system. The digital forensic readiness solutions include ‘Precognition’, which performs forensic analysis of suspected mobile applications. Precognition also uses machine learning techniques that utilize feature sets which are extracted from decompiled mobile applications, to identify potential security threats. The author has analyzed over 14151 mobile applications and classified vulnerabilities with an accuracy of 94.2%. The second solution, which concentrates on digital forensic readiness at the operating system level, securely preserves date and time stamps of targeted events running in smartphones, smartwatches, and wearables. These timestamps can be used to validate the digital evidence during a subsequent digital forensic investigation for any of the devices mentioned above. As a third contribution towards promoting digital forensic readiness in the mobile computing devices, the author has presented a novel form of forensic analysis technique, which analyzes over 5498 mobile ads to build the user profile which can be applied to identify the suspect who uses a particular device. The security contribution of the thesis includes an automated security analysis solution for identifying potential mobile security threats. Further, a solution ‘SecureRing’ has been proposed for securing the mobile applications to provide an additional layer of protection against attacks. The author has also designed ‘MobiSecureWrap’, which is an automated solution for wrapping mobile application binaries with additional security layer to protect them against potential threats. MobiSecureWrap recommends secure solutions based on detected security threats to protect the application binaries. The author evaluated over 5121 mobile applications to achieve a solution recommendation accuracy of 95.3%.
2018-05-01T00:00:00ZUnraveling representations for face recognition : from handcrafted to deep learning
http://repository.iiitd.edu.in/xmlui/handle/123456789/702
Unraveling representations for face recognition : from handcrafted to deep learning
Goswami, Gaurav; Singh, Richa (Advisor); Vatsa, Mayank (Advisor)
Automatic face recognition in unconstrained environments is a popular and challenging research problem. With the improvements in recognition algorithms, focus has shifted from addressing various covariates individually to performing face recognition in truly unconstrained scenarios. Face databases such as the YouTube Faces and the Point-and-shoot-challenge capture a wide array of challenges such as pose, expression, illumination, resolution, and occlusion simultaneously. In general, every face recognition algorithm relies on some form of feature extraction mechanism to succinctly represent the most important characteristics of face images so that machine learning techniques can successfully distinguish face images of one individual apart from those of others. This dissertation proposes novel feature extraction and fusion paradigms along with improvements to existing methodologies in order to address the challenge of unconstrained face recognition. In addition, it also presents a novel methodology to improve the robustness of such algorithms in a generalizable manner.
We begin with addressing the challenge of utilizing face data captured from consumer level RGB-D devices to improve face recognition performance without increasing the operational cost. The images captured using such devices is of poor quality compared to specialized 3D sensors. To solve this, we propose a novel feature descriptor based on the entropy of RGB-D faces along with the saliency feature obtained from a 2D face. Geometric facial attributes are also extracted from the depth image and face recognition is performed by fusing both the descriptor and attribute match scores. While score level fusion does increase the robustness of the overall framework, it cannot take into account and utilize the additional information present at the feature level. To address this challenge, we need a better feature-level fusion algorithm that can combine multiple
features while preserving as much of this information before the score computation stage. To accomplish this, we propose the Group Sparse Representation based Classifier (GSRC) which removes the requirement for a separate feature-level fusion mechanism and integrates multiple features seamlessly into classification. We also propose a kernelization based extension to the GSRC that further improves its ability to separate classes that have high inter-class similarity.
We next address the problem of efficiently using large amount of video data to perform face recognition. A single video contains hundreds of images, however, not all frames of a video contain useful features for face recognition and some frames might even deteriorate performance. Keeping this in mind, we propose a novel face verification algorithm which starts with selecting featurerich frames from a video sequence using discrete wavelet transform and entropy computation. Frame selection is followed by learning a joint representation from the proposed deep learning architecture which is a combination of stacked denoising sparse autoencoder and deep Boltzmann machine. A multilayer neural network is used as classifier to obtain the verification decision.
Currently, most of the highly accurate face recognition algorithms are based on deep learning based feature extraction. These networks have been shown in literature to be vulnerable to engineered adversarial attacks. We assess that non-learning based image-level distortions can also adversely affect the performance of such algorithms. We capitalize on how some of these errors propagate through the network to devise detection and mitigation methodologies that can help improve the real-world robustness of deep network based face recognition. The proposed algorithm does not require any re-training of the existing networks and is not specific to a particular type of network. We also evaluate the generalizability and efficacy of the approach by testing it with multiple networks and distortions. We observe favorable results that are consistently better than existing methodologies in all the test cases.
2018-11-01T00:00:00Z