Manjusha Pandey

Assessment of Object Detection Using Deep Convolutional Neural Networks

Detecting the objects from images and videos has always been the point of active research area for the applications of computer vision and artificial intelligence namely robotics, self-driving cars, automated video surveillance, crowd management, home automation and manufacturing industries, activity recognition systems, medical imaging, and biometrics. The recent years witnessed the boom of deep learning technology for its effective performance on image classification and detection challenges in visual recognition competitions like PASCAL VOC, Microsoft COCO, and ImageNet. Deep convolutional neural networks have provided promising results for object detection by alleviating the need for human expertise for manually handcrafting the features for extraction. It allows the model to learn automatically by letting the neural network to be trained on large-scale image data using powerful and robust GPUs in a parallel way, thus, reducing training time. This paper aims to highlight the state-of-the-art approaches based on the deep convolutional neural networks especially designed for object detection from images.

A Probabilistic Packet Filtering-Based Approach for Distributed Denial of Service Attack in Wireless Sensor Network

Wireless sensor networks (WSNs) are widely used networks that have lured attention of varied research fields due to their numerous ranges of applications. They have limited energy and power consumption, memory, communication, and computation capabilities. They are also distributed and randomly deployed. Due to the above-listed features, they are prone to various security threats and attacks. Distributed denial of service (DDoS) attack is one among them. These attacks aim at flooding the victim with abundant packets so as to exhaust its resources and cripple its capacity to receive desired packets and give its response accordingly. The network becomes congested and the victim becomes either unresponsive leading to denial of service or its response gets delayed. In this paper, we propose a mitigation mechanism that will curb the attempts of the attackers aiming to flood the WSN so as to cause denial of service with multitude of packets within a time span.

Deep Learning Approaches for Detecting Objects from Images: A Review

Detecting objects from images is a challenging problem in the domain of computer vision and plays a very crucial role for wide range of real-time applications. The ever-increasing growth of deep learning due to availability of large training data and powerful GPUs helped computer vision community to build commercial products and services which were not possible a decade ago. Deep learning architectures especially convolutional neural networks have achieved state-of-the-art performance on worldwide competitions for visual recognition like ILSVRC, PASCAL VOC. Deep learning techniques alleviate the need of human expertise from designing the handcrafted features and automatically learn the features. This resulted into use of deep architectures in many domains like computer vision (image classification, visual recognition) and natural language processing (language modeling, speech recognition). Object detection is one such promising area immensely needed to be used in automated applications like self-driving cars, robotics, drone image analysis. This paper analytically reviews state-of-the-art deep learning techniques based on convolutional neural networks for object detection.

Map-reduce based modeling and dynamics of infectious disease

The rapid increase in population creates an issue in handling and analyzing the population data for the traditional data base management system. So Big data came into figure to solve the issue. Big data is more efficient in comparison to the traditional data base system due to some of its basic features like Velocity, Veracity, Volume, Verity and Value. Day by day the disease are growing and becoming harmful to the society irrespective of treatments that are available. Infectious disease is caused by infectious agents including Viruses, Prions, Bacteria, Nematodes etc. Population dynamics is a branch of life science which includes the study of population size and age composition of dynamic system and the biological and environmental process managing them. This proposed paper consider an infectious disease i.e Dengue Fever and divides the population dynamic into three parts those are High Vulnerable, Mid vulnerable, Low vulnerable to Dengue. Then suggest the preventive measure like Forced preventive for high Vulnerable, Efficient preventive measure for Mid vulnerable and Delayed preventive measure for Low vulnerable areas by utilizing the benefits of big data.

Feedback analysis using big data tools

With the ever increasing man-machine interaction, automation of process and decline in hardware and software cost, the amount of digital data generated and used is increasing day by day. The big data referred here is the massive amount of digital data generated in each and every second in structured, semi-structured and unstructured format throughout the world. This emerging field of big data analytic has driven the researcher worldwide toward design, development and implementation of various tools, technologies, architecture and platforms for analyzing the huge volume of data generated day to day. Big data consist of data sets which is difficult for legacy database management system to analysis. This paper details some analysis like feedback analysis, sentiment analysis and word-count. Feedback are important for the system enhancement, finding loop holes and as well as for proper work distribution. Feedback is valuable information that will be used to make good decision. Feedback is important not only when it highlights weaknesses but also for strengths. If analysis of feedback is done in wrong way then the result of analysis will also be wrong. As a result, the pattern identified will also be incorrect thus making the whole system incorrect as a whole. We will be implementing this proposed system for feedback analysis using Map-Reduce framework for processing large data set and for storage we will use Hadoop.

Name node performance enlarging by aggregator based HADOOP framework

Due to the increased market competition increased data management and analysis has landed as in an era that requires further optimization data management and analysis. Big data technologies like apache HADOOP provide a frame work for parallel data processing and generation of analyzed results. MAPREDUCE method is used for analysis of data using various data analysis algorithms like clustering, fragmentation and aggregation. As per the HADOOP architecture the data received from client is distributed to various data node by the name node and it is the responsibility of name node to track the task being performed by a data nodes through a task-tracker, The presented proposal aims to reduce the burden on name node in the HADOOP architecture by providing the assistance through aggregator node which act as interface between the name node & data node.

Real time financial analysis using big data technologies

Due to advancement in Science and Technologies there are enormous amount of data available on internet. A large volume of structured, semi-structured and unstructured data is being created at a very rapid speed every day from heterogeneous sources like reviews, ratings, feedbacks, shopping details, etc., it is termed as Big Data. This data generated from different users share many common patterns which can be filtered and analysed to give some recommendation regarding the product, goods or services in which a user is interested. Recommendation systems are the software tools used to give suggestions to users on the basis of their requirements. Many people are not so much aware of different profitable and economical alternatives before using their money for goods or services. They are not so intelligent that they can quickly compare and judge that which product or service is better. The presented paper proposed a recommended system for management and utilisation of three components of salary i.e. saving, investment and expenditure. Many savings and investment consulting systems are available but no system provides effective and efficient recommendation regarding management and beneficial utilisation of salary. The advantage of proposed recommended system is that it provides better suggestion to a person for saving, expenditure and investment of their salary which in turns maximises their wealth. Due to enormous amount of data involved, Apache Hadoop framework is used for distributed processing. Apache Mahout is used for analysing the data and implementation of the recommender system.

Performance of elasticsearch in cloud environment with nGram and non-nGram indexing

The fact that technology have changed the lives of human beings cannot be denied. It has drastically reduced the effort needed to perform a particular task and has increased the productivity and efficiency. Computers especially have been playing a very important role in almost all fields in todays world. They are used to store large amount of data in almost all sectors, be it business and industrial sectors, personal lives or any other. The research areas of science and technology uses computers to solve complex and critical problems. Information is the most important requirement of each individual. In this era of quick-growing and huge data, it has become increasingly illogical to analyse it with the help of traditional techniques or relational databases. New big data instruments, architectures and designs have come into existence to give better support to the requirements of organizations/institutions in analysing large data. Specifically, Elasticsearch, a full-text java based search engine, designed keeping cloud environment in mind solves issues of scalability, search in real time, and efficiency that relational databases were not able to address. In this paper, we present our own experience with Elasticsearch an open source, Apache Lucene based, full-text search engine that provides near real-time search ability, as well as a RESTful API for the ease of user in the field of research.

Improvising name node performance by aggregator aided HADOOP framework

Due to the increased market competition increased data management and analysis has landed as in an era that requires further optimization data management and analysis. Big data technologies like apache HADOOP provide a frame work for parallel data processing and generation of analyzed results. MAPREDUCE method is used for analysis of data using various data analysis algorithms like clustering, fragmentation and aggregation. As per the HADOOP architecture the data received from client is distributed to various data node by the name node and it is the responsibility of name node to track the task being performed by a data nodes through a task-tracker, The presented proposal for “Improvising Name Node Performance By Aggregator Aided HADOOP Framework” aims to reduce the burden on name node in the HADOOP architecture by providing the assistance through aggregator node which act as interface between the name node & data node.

Concealing of the base station's location for preserving privacy in Wireless Sensor Network by mitigating traffic patterns

A Wireless Sensor Network (WSN) actually consists of a huge amount of small and tiny individual sensor nodes which are well capable of sensing and collecting data samples from the environment and are treated as the information source. These sensor nodes helps in monitoring physical or environmental conditions and also pass their data samples to the main location through the network. In wireless sensor networks the central point of failure is a base station which collects data from several intermediate nodes. But problem occurs when these intermediate nodes fail to forward the data to the destination node. Attacker examines the packet traffic and can easily reach the base station to destroy it thereby rendering the entire sensor network inoperative. In this paper the proposed privacy preservation mechanism mainly focuses on minimizing the energy consumption thereby increasing the network life time and hence reduces the communication overhead in the network as a whole. Hence our proposed scheme involves generation of fake packets by the nodes considering remaining energy of each node.