The use of electroencephalography (EEG) for bio-metric recognition, in particular for verification systems, has increased in the last years, due to some advantages that EEG signals present when used as signatures, as compared to other identifiers. In this paper we explore the use of unsupervised adaptive learning as a tool for enhancing the features representing each possible subject in a biometric system, in order to improve its performance. To do so, we designed three different frameworks based on Self Organizing Maps (SOM) neural networks, and compared their performance with a base model using no enhancement. Our experiments, using different input tasks and two combinations in the number of channels, with data obtained from two public EEG databases, showed that a SOM with Dynamic Structure (GSOM) obtained the best Equal Error Rate (EER). Such EER was 0.08 ± 0.04 when using as input the counting task of a public database provided by the University of Colorado, and an EER of 0.11 ± 0.04 was obtained for the rotation task in the same database. We also assessed our frameworks using the public database BIOMEXDB, provided by INAOE, where we also found that GSOM outperformed other state-of-the-art works.
Cancer is a disease in which the body cells start growing uncontrollably and spreads all over the body. Mostly, the cancer symptoms appear only in the advanced stages. This disease is very complex in terms of its diagnosis in the early stages which results in a high mortality rate. Thus, there is a requirement for cancer to be diagnosed at its early stages which may result in better survival chances and the patients can be treated successfully. The dose-limiting toxicity in lung cancer radiotherapy (RT) is radiation pneumonitis (RP). Cancer characteristics and treatment features are intertwined, resulting, in RP associated with a single parameter is not always possible. This study aims to determine the algorithms which are most accurate for lung cancer prediction. As per the study by WHO, it has been found that in the year 2020, a total of 2.21 million people were diseased with lung cancer resulting in 1.80 million deaths all over the globe. In India, each year almost 70,000 active cases of lung cancer are identified. Early detection plays an important role in saving lives because it can give a patient a better chance to cure and recover. In recent times, different computer technologies are used for solving the problems of cancer detection. In this work, several types of machine-learning algorithms such as Naive Bayes (accuracy 96.61%), Decision tree (accuracy 91.52%), Random forest (accuracy 93.22%), Logistic Regression (accuracy 96.61%), Multilayer perceptron (accuracy 98.30%) have been utilized for predicting lung cancer. Among all of these algorithms, multilayer perceptron is the best algorithm to diagnose lung cancer.
This paper presents a four staged voltage-controlled ring oscillator (VCO) operating in wide range of frequency that consumes low power, improves the performance and provides stability to the circuit. Due to the limitation of MOSFET that they cannot be scaled down after certain range due to some of its limitations such as high power and leakage current. To overcome the drawback, we used CNTFET technology that uses Carbon Nano Tubes in place of silicon. The VCO that is made in this paper using CNTFET operates in Terahertz frequency range varying from 0.331 THz to 0.091 THz and similarly the power dissipated ranges from 0.9565 to 0.2506 mW for control Voltage of 0 to 1 volts. The proposed VCO shows 97.7% improvement in power dissipation 99.4% increase in the frequency range in comparison to the VCOs based on CMOS technology. VCO produces a sinusoidal waveform and we checked the correctness of our design by verifying the waveform produced and simulated results.
Life expectancy (LE) models provide a lot of ways to improve healthcare and other social welfares related to society. Life expectancy models provide solutions to problems like how to decide on retirement age or manage financial issues related to public matters. These models are becoming prominent in many regions as they are being widely used by government bodies and private sector for their policy making and developing health integrated systems. Thus, this paper aims to analyze the Trends in Life Expectancy in about 72 countries of the world over a span of 16 years, i.e., from 2000-2015. The study gives plots of attributes such as life expectancy, GDP, infant deaths, adult mortality, etc. across year which would help the countries understand the life expectancy trends over the course of time and suggest areas which should be focused upon to efficiently increase the life expectancy of its population. The simulations are done in Google Collab by using various Python libraries like pandas, numpy, matplotlib (used for plotting graphs), seaborn (used for plotting 3-D graphs and advanced visualization features of python), sklearn (used for handling missing data), and plotly express (used for plotting choropleth).
Recognizing human movements through computer vision is an important field of research, which can be used in various applications such as patient monitoring, observation, and human-machine interface. The ability to perceive these movements requires extremely complex judgments. Generally, the above-mentioned applications need to automatically recognize advanced operations, such as: a pair of easy movements of a man and a woman. If the action is well classified, then the proper information can be provided to the system. This paper addresses various machine learning algorithms such as logistic regression, RBF SVM, decision tree, random forest, linear SVM, gradient boosting DT by grouping different activities. This article classifies complex human behaviors by observing, comparing and evaluating the performance of algorithms by using large set of information.
