Showing papers by "Anupam Shukla published in 2018"

Three-dimensional path planning for unmanned aerial vehicles using glowworm swarm optimization algorithm

Abstract: Robot path planning is a task to determine the most viable path between a source and destination while preventing collisions in the underlying environment. This task has always been characterized as a high dimensional optimization problem and is considered NP-Hard. There have been several algorithms proposed which give solutions to path planning problem in deterministic and non-deterministic ways. The problem, however, is open to new algorithms that have potential to obtain better quality solutions with less time complexity. The paper presents a new approach to solving the 3-dimensional path planning problem for a flying vehicle whose task is to generate a viable trajectory for a source point to the destination point keeping a safe distance from the obstacles present in the way. A new algorithm based on discrete glowworm swarm optimization algorithm is applied to the problem. The modified algorithm is then compared with Dijkstra and meta-heuristic algorithms like PSO, IBA and BBO algorithm and their performance is compared to the path optimization problem.

Multi-robot multi-target dynamic path planning using artificial bee colony and evolutionary programming in unknown environment

Abstract: Navigation or path planning is the basic need for movement of robots. Navigation consists of two foremost concerns, target tracking and hindrance avoidance. Hindrance avoidance is the way to accomplish the task without clashing with intermediate hindrances. In this paper, an evolutionary scheme to solve the multi-agent, multi-target navigation problem in an unknown dynamic environment is proposed. The strategy is a combination of modified artificial bee colony for neighborhood search planner and evolutionary programming to smoothen the resulting intermediate feasible path. The proposed strategy has been tested against navigation performances on a collection of benchmark maps for A* algorithm, particle swarm optimization with clustering-based distribution factor, genetic algorithm and rapidly-exploring random trees for path planning. Navigation effectiveness has been measured by smoothness of feasible paths, path length, number of nodes traversed and algorithm execution time. Results show that the proposed method gives good results in comparison to others.

An intelligible deep convolution neural network based approach for classification of diabetic retinopathy

Abstract: Abstract Deep convolution neural networks (CNNs) have demonstrated their capabilities in modern-day medical image classification and analysis. The vital edge of deep CNN over other techniques is their ability to train without expert knowledge. Time bound detection is very beneficial for the early cure of disease. In this paper, a deep CNN architecture is proposed to classify nondiabetic retinopathy and diabetic retinopathy fundus eye images. Kaggle 2015 diabetic retinopathy competition dataset and messier experiment dataset are used in this study. The proposed deep CNN algorithm produces significant results with 93% area under the curve (AUC) for the Kaggle dataset and 91% AUC for the Messidor dataset. The sensitivity and specificity for the Kaggle dataset are 90.22% and 85.13%, respectively; the corresponding values of the Messidor dataset are 91.07% and 80.23%, respectively. The results outperformed many existing studies. The present architecture is a promising tool for diabetic retinopathy image classification.

Mathematical analysis of schema survival for genetic algorithms having dual mutation

Abstract: Genetic algorithms are widely used in the field of optimization. Schema theory forms the foundational basis for the success of genetic algorithms. Traditional genetic algorithms involve only a single mutation phase per iteration of the algorithm. In this paper, a novel concept of genetic algorithms involving two mutation steps per iteration is proposed. The purpose of adding a second mutation phase is to improve the explorative power of the genetic algorithms. All the possible cases regarding the working of the proposed variant of the genetic algorithms are explored. After a meticulous analysis of all these cases, three lemmas are proposed regarding the survival of a schema after the application of the dual mutation. Based on these three lemmas, a theorem is proved, and a mathematical expression representing the probability of survival of a schema after the application of the crossover and dual mutation is derived. This expression provides a new insight about the penetration of a schema for such scenario and improves our understanding of the functioning of this modified form of the genetic algorithm.

Deep Learning Based Unsupervised POS Tagging for Sanskrit

Abstract: In this paper, we present a deep learning based approach to assign POS tags to words in a piece of text given to it as input. We propose an unsupervised approach owing to the lack of a large Sanskrit annotated corpora and use the untagged Sanskrit Corpus prepared by JNU for our purpose. The only tagged corpora for Sanskrit is created by JNU which has 115,000 words which are not sufficient to apply supervised deep learning approaches. For the tag assignment purpose and determining model accuracy, we utilize this tagged corpus. We explore various methods through which each Sanskrit word can be represented as a point multi-dimensional vector space whose position accurately captures its meaning and semantic information associated with it. We also explore other data sources to improve performance and robustness of the vector representations. We use these rich vector representations and explore autoencoder based approaches for dimensionality reduction to compress these into encodings which are suitable for clustering in the vector space. We experiment with different dimensions of these compressed representations and present one which was found to offer the best clustering performance. For modelling the sequence in order to preserve the semantic information we feed these embeddings to a bidirectional LSTM autoencoder. We assign a POS tag to each of the obtained clusters and produce our result by testing the model on the tagged corpus.

Machine Translation : From Statistical to modern Deep-learning practices.

Abstract: Machine translation (MT) is an area of study in Natural Language processing which deals with the automatic translation of human language, from one language to another by the computer. Having a rich research history spanning nearly three decades, Machine translation is one of the most sought after area of research in the linguistics and computational community. In this paper, we investigate the models based on deep learning that have achieved substantial progress in recent years and becoming the prominent method in MT. We shall discuss the two main deep-learning based Machine Translation methods, one at component or domain level which leverages deep learning models to enhance the efficacy of Statistical Machine Translation (SMT) and end-to-end deep learning models in MT which uses neural networks to find correspondence between the source and target languages using the encoder-decoder architecture. We conclude this paper by providing a time line of the major research problems solved by the researchers and also provide a comprehensive overview of present areas of research in Neural Machine Translation.

A new insight into the schema survival after crossover and mutation for genetic algorithms having distributed population set

Abstract: Genetic algorithms are one of the most popular optimization algorithms. Schema theory provides a mathematical foundation for the working of genetic algorithm. Different variants of the basic genetic algorithm have been proposed; and genetic algorithm having distributed population set (Island model of genetic algorithm) is one of them. In this paper, traditional schema theory is extended for the case of genetic algorithm having distributed population set and the probability of schema survival for such case is derived.

Discrete bacteria foraging optimization algorithm for graph based problems - a transition from continuous to discrete

Abstract: Bacteria Foraging Optimisation Algorithm is a collective behaviour-based meta-heuristics searching depending on the social influence of the bacteria co-agents in the search space of the problem. Th...

Human-Machine Interface System for pre-Diagnosis of Diseasesusing Machine Learning

Abstract: The rapid growth of applications of latest information technology into the field of medical sciences have founded the idea to develop such a platform through which pre-diagnosis of diseases could be easy, efficient and less time consuming. This paper talks about two frameworks designed using machine learning algorithms such as ANN, SVM and Decision Tree Induction to develop the models through which a number of diseases can be pre-diagnosed simultaneously with the analysis of symptoms initially recorded in the patient's body. These symptoms and physical readings have been taken as inputs to produce the output i.e. the predicted disease. The most important factors contributing for multiple disease prediction were determined such as age, sex, body temperature, blood pressure and symptoms like nausea, vomiting and fever. Data sets were collected from different hospitals in India during this research. All the models used were able to perform with an accuracy above 85%.