A tutorial on the design and development of automatic speaker-recognition systems is presented. Automatic speaker recognition is the use of a machine to recognize a person from a spoken phrase. These systems can operate in two modes: to identify a particular person or to verify a person's claimed identity. Speech processing and the basic components of automatic speaker-recognition systems are shown and design tradeoffs are discussed. Then, a new automatic speaker-recognition system is given. This recognizer performs with 98.9% correct identification. Last, the performances of various systems are compared.

Speaker recognition : A tutorial : Automated biometrics

Traffic of non-congestion control on the internet is on rise, traffic varies from TCP traffic to congestion collapse. Router mechanisms need to be included to find and stop congestion in the time of data flow and encourage end-to-end congestion in architecture for future work. Different approaches have been scrutinized to manage and control the data flow to say the correct flow in bandwidth regulations. End to end congestion control now has few problems such as congestion collapse from the packets which are not delivered. The end-to-end factor is very important in congestion control but just congestion control algorithms alone cannot stop congestion collapse and are not effective in giving response during crowding in the network. So NBP (Network Border Patrol), which is a mechanism used to avoid congestion in a network, has been used. A leaky bucket has been used to monitor the flow in per-flow rate.

Fault tolerance using Leakybucket and Network Border Patrol

The security of modern IoT Industry 4.0, 5G, 6G, Mobile ad hoc (MANET), narrowband internet of things (NB-IoT) and wireless sensory (WSN) networks and the autonomous computing capabilities of individual devices and self-organizing, greatly influence their applications in smart connected world. To achieve the sufficient security and privacy, autonomous and dynamic adaptive key management scheme and mutual authentication protocols should be provided to validate legitimacies of large number of nodes, users, devices, and servers in 5G networks. To this end, we propose a lightweight cipher approach that enhances the security level of communications and enables authentication and access control scheme that is implemented on a multi-layer communication architecture designed for NB-IoT 5G networks and resolve heterogeneity and scalability issues. It is well known that establishing an efficient cryptographic key generation and management scheme for machine-to-machine (M2M) and device-to-device (D2D) communication for IoT networks in 5G, the purposes of encryption and decryption is severely constrained by computation resources and performance. As a part of the network security, the key management plays a vital role in it. When compared to other schemes in PKI such as public key, pairwise-key and group-based key management, the dynamic-key based security is efficient in resource utilization and scalable which are the two key performance factors for Cloud based IoT/wireless sensory networks. Usually most IoT networks use a less secure communication channel that employs session keys in order to encrypt the communication. Additionally, in most IoT networks the resource consumption is constrained which then results in existing dynamic key generation algorithms being infeasible to run. It is still a challenging problem to securely interconnect and operate IoT protocols between constrained embedded devices. We propose a novel dynamic key generation scheme that takes the entropy and performs various operations to continuously generate a large set of unique keys. This type of key generation model is predominantly suitable for the conditions where the IoT devices cannot rely on constant key negotiation with dedicated servers as well as those devices which cannot reuse existing keys for encryption purposes. Performance evaluation using simulation and case study demonstrates that our dynamic key establishment scheme ensures an enhanced security level while reducing the communication overhead and the average latency for different category of IoT applications in 5G networks.

A lightweight authentication scheme for 5G mobile communications: a dynamic key approach

In order to identify cattle breeds according to their size and color, Cattle Identification techniques are used. Present techniques used for individual cattle breed prediction needs unique radio frequency, visual frequency or tagging of ear. To identify a cattle breed, we suggest identification using several image datasets of cattle’s. An image of the cattle will be taken to get the prominent features of the front, rear and side view and these extracted features will be taken to identify each individual cattle with their information. Each image dataset will be processed and according to the processed image the cattle breed will be determined. Each cattle image will be pre-processed and after pre-processing of the images they will be categorized into their respective breed. A separate category of rare breeds of cattle’s will be categorized using the pre-processed image dataset of the cattle breeds. CNN deep learning method is used in this work to recognize the face and different shapes in the whole body of the cattle image.

Identification of Cattle Breed using the Convolutional Neural Network

Research methods in machine learning play a pivotal role since the accuracy and reliability of the results are influenced by the research methods used. The main aims of this paper were to explore current research methods in machine learning, emerging themes, and the implications of those themes in machine learning research.  To achieve this the researchers analyzed a total of 100 articles published since 2019 in IEEE journals. This study revealed that Machine learning uses quantitative research methods with experimental research design being the de facto research approach. The study also revealed that researchers nowadays use more than one algorithm to address a problem. Optimal feature selection has also emerged to be a key thing that researchers are using to optimize the performance of Machine learning algorithms. Confusion matrix and its derivatives are still the main ways used to evaluate the performance of algorithms, although researchers are now also considering the processing time taken by an algorithm to execute. Python programming languages together with its libraries are the most used tools in creating, training, and testing models. The most used algorithms in addressing both classification and prediction problems are; Naive Bayes, Support Vector Machine, Random Forest, Artificial Neural Networks, and Decision Tree. The recurring themes identified in this study are likely to open new frontiers in Machine learning research.

/pdf/research-methods-in-machine-learning-a-content-analysis-5bkokmnc86.pdf

Research Methods in Machine Learning: A Content Analysis

The shortage of water around the world force us to minimize the usage of water. More than 75% of fresh water resources were using for irrigation purpose so efficient utilization of water in irrigation system with advanced method is required. This paper presents an advanced technology based smart system to predict the irrigation requirements of a field by sensing of ground parameter like moisture of the soil, temperature-humidity and water level using Machine learning algorithm (ML). Existing algorithms like K-mean and SVM are facing overfitting problem to overcome this problem we are using K-Nearest neighbor method along with this proposed system has a capacity to realize a fully automated irrigation scheme and discussed in detail about the information processing results with three weeks pre-defined data based on the proposed algorithm. The system is capable and the prediction results are more accurate.

Efficient Machine Learning Algorithm for Smart Irrigation

Diabetic retinopathy is an ailment of the retinal vasculature that ultimately develops to some diploma in nearly all patients with lengthy-status diabetes Proliferative diabetic retinopathy is an uncommon circumstance in all likelihood to cause acute visual deficiency It is observed via the growth of unusual new retinal vessels To symbolize the improvement of irregular new retinal vessels, an algorithm for spontaneously identifying new vessels on the optic disc using retinal photographs is described The algorithm takes Five module method (FMM) compressed retinal images as the input Watershed lines and canny detectors are used to find the vessel like candidate segment Different features namely shape of the segment, position of the segment from the origin, positioning, intensity of the segment in the image, divergence, and line density are extracted for each candidate segment Each candidate segment is labeled as normal or abnormal based on its features using Support Vector Machine (SVM) classifier The experimentation results suggests that the automated retinopathy analysis system provides clinical insights in detecting the ailment

Segmentation and classification of FMM compressed retinal images using watershed and canny segmentation and support vector machine

In today's accessible world, web has turned into a fundamental requirement of everybody. Barring the principal choices of the web, i.e., aquatics and net perusing, we'd like further choices like diversion, on-line video spilling and meeting up and so on. To use entirely unexpected alternatives of the web, very surprising gadgets are offered that changes as far as esteem, figuring force, and show and vitality consumption. Amid this paper, we tend to blessing however entirely unexpected alternatives of the web might be surfed by having one gadget Raspberry Pi. We tend to think about very surprising attributes of Raspberry Pi with in vogue processing gadgets like tablet PC, Smartphones convenient PC and TV. Our outcomes demonstrate that basic registering assignments might be expert in extra vitality practical means. We have a tendency to demonstrate that Raspberry Pi expends less power and may save eminent amount of vitality in routine processing undertakings.  

/pdf/energy-and-performance-analysis-of-raspberry-pi-with-modern-3nm6gdeiid.pdf

Energy and Performance Analysis of Raspberry Pi with Modern Computing Devices

A survey on classification and clustering algorithms for uncompressed and compressed text

Object tracking is a procedure of movement estimation in computer vision application we present a combined multiple object tracking technique for a video. A video is a frame by frame sequence of images. Optical flow is a flexible representation of visual motion that is particularly suitable for computers for analyzing digital images. In this work the Horn & Schunck method is used to find the optical flow vectors which in turn pave a way for the detection and tracking of the single moving object in a video. Kalman filter removes the noise that effects a background subtracted image and predicts the position of an object accurately. A combination of Optical flow and Kalman filter method is designed in order to attain an accurate object tracking system. The accuracy of occluded object in dynamic background is promising compared to simple background subtraction. The experiments are conducted on different videos to prove the efficiency and the results are discussed. KeywordKalman Filters (KF), Optical flow (OF), Background Subtraction, Occlusion Handling.

S. Akshay

Papers

Efficient Machine Learning Algorithm for Smart Irrigation

Segmentation and classification of FMM compressed retinal images using watershed and canny segmentation and support vector machine

Energy and Performance Analysis of Raspberry Pi with Modern Computing Devices

A survey on classification and clustering algorithms for uncompressed and compressed text

Improved multiple object detection and tracking using KF-OF method