6^th International Conference on Artificial Intelligence and Applications (AI 2020)

Zero-Visibility Cops And Robber Game On The Cube Grid

Jiahui Wang and Farong Zhong, Department of Computer Science, Zhejiang Normal University, Jinhua, China

ABSTRACT

Cops and robber game is a graph searching problem. The robber is invisible in the zero-visibility cops and robber game. In this paper, we study the zero-visibility cops and robber game on the cube grid. We first study a partition problem of the cube grid . Then we prove the lower bound on the zero-visibility cop number of the cube grid by using the results in the partition. We also show the relationship between the zero-visibility cop number and the parameter of .

KEYWORDS

Cops and robber, Partition, Boundary, Zero-visibility Cop number.

Performance Analysis of Machine Learning Algorithms for Intrusion Detection using UNSW-NB15 dataset

Geeta Kocher¹ and Gulshan Kumar², ¹Research Scholar, Department of Computational Sciences, MRSPTU, Bathinda, Punjab, India, ²Assosiate Professor, Department of Computer Applications, SBSSTC, Ferozpur, Punjab

ABSTRACT

With the advancement of internet technology, the numbers of threats are also rising exponentially. To reduce the impact of these threats, researchers have proposed many solutions for intrusion detection. In the literature, various machine learning classifiers are trained on older datasets for intrusion detection which limits their detection accuracy. So, there is a need to train the machine learning classifiers on latest dataset. In this paper, UNSW-NB15, the latest dataset is used to train machine learning classifiers. On the basis of theoretical analysis, taxonomy is proposed in terms of lazy and eager learners. From this proposed taxonomy, K-Nearest Neighbors (KNN), Stochastic Gradient Descent (SGD), Decision Tree (DT), Random Forest (RF), Logistic Regression (LR) and Naïve Bayes (NB) classifiers are selected for training. The performance of these classifiers is tested in terms of Accuracy, MSE, Precision, Recall, F1-Score, TPR and FPR on UNSW-NB15 dataset and comparative analysis of these machine learning classifiers is carried out.

KEYWORDS

Intrusion Detection System, Random Forest, KNN, UNSW-NB15, Machine Learning Algorithms.

Multi Image Steganography Using Distributed LSB Algorithm and Secret Text Recovery on Stego Image Corruption

Jagan Raj Jayapandiyan¹, C. Kavitha² and K. Sakthivel³, ¹Periyar University, India, ²Thiruvalluvar Govt. Arts College, India, ³K. S. Rangasamy College of Technology, India

ABSTRACT

In this proposed research work, an attempt has been made to use multiple image files for steganography encoding along with the capability of secret text recovery in the event of any image corruption during the transit. This algorithm is effective on the security factor of secret image since the embedded checksum will validate for any unauthorized users or intruders attempt to corrupt the picture in any aspect. If any of the stego image underwent any steganalysis or MiM attack, then this proposed algorithm can effectively regenerate the content of one stego image using other intact stego images received in the receiving end.

KEYWORDS

Steganography, Multi-cover image, secret message recovery.

Content based Remote Sensing Image Retrieval using Artificial Neural Network

Aswathi L P and Anoop K, Department of Electronics and communication Engineering, College of Engineering Thalassery, Kannur, Kerala, India

ABSTRACT

Remote sensing is being used in different fields like agriculture,research etc..Remote sensed images contains complex visual contents. This paper explains about the content based remote sensing image retrieval using ANN. In remote sensing method the sensors which will be fixed on an aircraft or satellite is used for capturing remote sensing images. Due to the increase in the use of remote sensing technology and also the number of satellites used,the volume of image dataset is increasing exponentially. Content Based Remote sensing Image Retrieval is used as to reduce the difficult in managing large volume of earth data.

KEYWORDS

ANN, Remote sensing, GLCM.

Design of efficient dual character paradigm based P300 speller system and performing spectral analysis of EEG signals

Perla Abhinay Sai and Praneeth Sai, Electrical Engineering, National Institute Of Technology Raipur, India

ABSTRACT

Brain-computer interfaces (BCI) are systems that allow communication between the brain and computer. Their working basically consists of three sections: collecting the brain signal, interpreting the brain signal and in the end sending the commands to a connected machine in accordance to the brain signal received. Instead of depending on peripheral nerves and muscles, a BCI directly measures brain activity associated with the user’s intent and translates the recorded brain activity into corresponding control signals for BCI applications. An event-related potential, which is generated during the process of decision making is called as P300 wave as it has a positive peak at 300ms after visual stimulation. This report aims to detect the P300 wave from acquired EEG signal accurately. The existing R/C paradigms suffer from various problems such as crowding effect due large size as well as error related potential generation due to counting by subjects. Therefore, we came up with an idea to use dual character R/C paradigm which address the above issues by reducing size and using voluntary eye closure instead of counting. This will in turn lead to a significant increase in character detection. Therefore, people with severe motor neuron disability, suffering from amyotrophic lateral sclerosis (ALS) or brain injuries will be able to communicate with the people around them with ease.

KEYWORDS

Brain computer interface, Electroencephalography, fisher distance, k nearest neighbour.

Biometric Fetal Contour Extraction using Hybrid Level Set

Rachana Jaiswal and Srikant Satarkar, Department of Computer Science & Engineering, Amravati University, Akola, MH

ABSTRACT

In medical imaging, accurate anatomical structure extraction is important for diagnosis and therapeutic interventional planning. So, for easier, quicker and accurate diagnosis of medical images, image processing technologies may be employed in analysis and feature extraction of medical images. In this paper, some modifications to level set algorithm are made and modified algorithm is used for extracting contour of objects in an image. The proposed approach is applied on fetal ultrasound images. In traditional approach, fetal parameters are extracted manually from ultrasound images. Due to lack of consistency and accuracy of manual measurements, an automatic technique is highly desirable to obtain fetal biometric measurements. This proposed approach is based on global & local region information for fetal contour extraction from ultrasonic images. The primary goal of this research is to provide a new methodology to aid the analysis and feature extraction from fetal images.

KEYWORDS

Active contour, Region-based, Edge-based.

Segmentation of Dermatoscopic images using wavelet Transform

Kanchan A. Nandedkar, MGM’s college of Computer Science & Information Technology, Nanded-431605 (Maharashtra)

ABSTRACT

Wavelet based segmentation provides a very effective technique for medical images. Automatic segmentation of skin lesions is the first step towards development of computer aided diagnosis of melanoma. This paper provides an improved automated skin lesion segmentation method for color dermoscopic images. A novel wavelet transform based technique called “Segmentation of Dermatoscopic images using wavelet transform” (SDIWT) is proposed. One of the important advantages of wavelet transform is that it provides a precise and unifying framework for the analysis and characterization of a signal at different levels. SDIWT uses a perceptually uniform color space for segmentation. To reduce computational complexity for clustering, prominent pixels are selected. One level decomposition of Daubechies wavelet function is used. LL1 sub band of decomposition is utilized for clustering. Fuzzy c- means (FCM) clustering technique is used to find out clusters and their labels. Fuzzy entropy is used to decide number of clusters. The image pixels are classified to respective clusters based on minimum Euclidean distance. A post processing noise filtering stage is applied to improve the segmentation output. Edges are determined to specify the lesion from Dermatoscopic images. One of the advantages of the proposed method is that it does not require to specify a priori information to segment a color region. The competence of the proposed method has been demonstrated by various experiments.

KEYWORDS

Dermatoscopic segmentation, Fuzzy c- means, wavelet transform, lesion segmentation.

Image Processing Failure and Deep Learning Success in Lawn Measurement

J. Wilkins¹, M. V. Nguyen¹ and B. Rahmani^{1, 2}, ¹Fontbonne University, USA, ²Maryville University, USA

ABSTRACT

Lawn area measurement is an application of image processing and deep learning. Researchers used hierarchical networks, segmented images, and other methods to measure the lawn area. Methods’ effectiveness and accuracy varies. In this project, image processing and deep learning methods were used to find the best way to measure the lawn area. Three image processing methods using OpenCV compared to convolutional neural network, which is one of the most famous, and effective deep learning methods. We used Keras and TensorFlow to estimate the lawn area. Convolutional neural network or shortly CNN shows very high accuracy (94-97%). In image processing methods, thresholding with 80-87% accuracy and edge detection are the most effective methods to measure the lawn area while the method ofcontouring with 26-31% accuracy does not calculate the lawn area successfully. We may conclude that deep learning methods, especially CNN, could be the best detective method comparing to image processing learning techniques.

KEYWORDS

Lawn Measurement, Convolutional Neural Network, Thresholding, Edge Detection, Contouring.

Promises and Challenges of Reinforcement Learning Applications in Motion Planning of Automated Vehicles

Nikodem Pankiewicz, Tomasz Wrona, Wojciech Turlej and Mateusz Orlowski, AGH University of Science and Technology, Aptiv, Krakow, Poland

ABSTRACT

As automated driving development progresses forward, novel methods are required to handle the vastness of possible road situations and to face end user's high demands. Trying to solve the problem of motion control involving decision making and trajectory planning it is reasonable to take into consideration reinforcement learning as a viable approach. In this paper, we present the promises reinforcement learning can bring to automated driving domain and the list of challenges we encountered during our work. We address the issues related to the environment definition, sample efficiency, safety and explainability.

KEYWORDS

Reinforcement Learning, Automated Vehicle, Behaviour Planning, Decision Making.

Product Recommendation using Object Detection from Video, Based on Facial Emotion

Kshitiz Badola¹, Ajay Joshi² and Deepesh Sengar³, ¹Department of Computer Science, Mahavir Swami Institute of Technology, Guru Gobind Singh Indraprastha University, New Delhi, India, ²Department of Electronics, Deen Dayal Upadhyaya College (University of Delhi), New Delhi, India, ³Department of Computer Science, Riga Technical University, Riga, Latvia

ABSTRACT

In today’s world, with the increasing demand of products and their growing productivity from producers, customers sometimes failed to decide whether they are interested in buying a particular product or not. So author, here proposed a framework which deals with the buying of only items of interest, for a consumer. In our feature-set, whenever any consumer tends to watch any video from YouTube, it results in breakdown into several frames (frames per second), and from there we use object detection technique to detect each and every object in a particular frame, and then to find whether our consumer is interested in that particular object or not, we use facial emotion detector to check whether our user is happy, surprised, neutral or any other emotion. After viewing those products which are present in a frame of a video. Merging only those items of interest which were tend to fall for consumer’s positive choices (emotions), we then used Amazon online marketing technique to recommend products selected by our framework.

KEYWORDS

Convolutional Neural Networks, Facial Expressions, Object Detection, ImageAI, Selenium, Machine Learning.

Clinical Assessment and Management of COVID 19 Patients using Artificial Intelligence

Rashmi Phalnikar¹, Subhal Dixit² and Harsha Talele³, ¹Associate Professor, SCET, MIT WPU, Pune, ²Director ICU, Sanjeevan Hospital, Pune, ³Assistant Professor, KCES’s COEM, Jalgaon

ABSTRACT

The COVID-19 infection caused by Novel Corona Virus has been declared a pandemic and a public health emergency of international concern. Infections caused by Corona Virus have been previously recognized in people and is known to cause Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS). Unlike the earlier infections, COVID-19 spreads alarmingly and the experience and volume of the scientific knowledge on the virus is small and lacks substantiation. To manage this crisis, Artificial intelligence (AI) promises to play a key role in understanding and addressing the COVID-19 crisis. It tends to be valuable to identify the infection, analyse it, treat it and also predict the stages of infection.Artificial intelligence algorithms can be applied to make diagnosis of COVID-19 and stepping up research and therapy. The paper explains a detailed flowchart of COVID-19 patient and discusses the use of AI at various stages. The preliminary contribution of the paper is in identifying the stages where the use of Artificial Intelligence and its allied fields can help in managing COVID-19 patient and paves a road for systematic research in future.

KEYWORDS

Artificial Intelligence, COVID-19.

Application of Augmented reality & virtual reality in Architecture and Planning: an overview

Pearl Jishtu¹, Madhura A Yadav², ¹Student of Fifth year School of Architecture & Design, Manipal University Jaipur Dehmi Kalan, Off Jaipur-Ajmer Expressway, Jaipur - 303 007, Rajasthan, India, ² Professor Manipal University Jaipur Dehmi Kalan, Off Jaipur-Ajmer Expressway, Jaipur – 303 007, Rajasthan, India

ABSTRACT

AR and VR – simulation tools created to assist global evolution for saving time. Time as resource is difficult to harness; however, it would make work highly efficient and productive when tackled with automation. All concerned are excited about AR and VR’s involvement in our lifestyle, but not all have comprehended its impact. AR and VR in Architecture & Planning were introduced as assisting tools and has helped generate multiple design options, expanded possibilities of visualization, and provided us with more enhanced, detailed, and specific experience in real-time; enabling us to visualize the result of work at hand well before the commencement of the project. These tools are further being developed for city development decisions, helping citizens interact with local authorities, access public services, and plan their commute. After reviewing multiple research papers on AI, it was observed that all are moving forward with the changes brought by it, without entirely understanding its role. This paper provides an overview of the application of AR & VR in architecture and planning.

KEYWORDS

Virtual reality (VR), Augmented reality (AR), Architecture, Urban Design, Human Computer Interface (HCI)