The aim of human-computer interaction (HCI) is to create truly effective and beneficial interfaces using today's innovative technology. A well-designed interface is especially important to users. The development of HCI to the present is highlighted and the chapter briefly describes HCI's primary components, both human and computer. The various challenges of HCI design, its characteristics, and principles are examined, as well as several HCI guidelines, proposed by experts. The chapter explains general design principles as well as their testing techniques. HCI devices have been developed into recent technologies, such as eye-tracking, or speech recognition. The Internet of Things, cloud computing, and its tools are discussed. Finally, the applications of HCI as well as their advantages and disadvantages are examined.
Gastric cancer is one of the most commonly occurring cancers in the world, averaging about 1.9 million cases in 2022 alone. Gastric cancer is observed to be twice as common in men than in women. Early diagnosis is absolutely crucial in identifying gastric cancer. The gene expression dataset GSE2685 used in this study was collected from the Gene Expression Omnibus (GEO) which consists of 4524 features and 30 samples. Since the dataset consists of numerous features, the feature selection is employed using Analysis of Variance (ANOVA) test to select the top 50 features which can be used for modeling. Machine learning algorithms are used to predict the presence of cancer in patients using historical patient data. In this study, different machine learning techniques are used to identify and classify the type of gastric intestinal (GI) cancer. The study demonstrates that the decision tree (DT) model was the best model with the highest accuracy.
Predicting stock prices and index movement in the field of finance is always challenging. The events in the macro-economic framework affect the trends of the market and the COVID-19 pandemic was a major reason for the slowdown of the global economies in the short run. It was assumed that the healthcare industry has completely been transformed due to changing behavioral habits of individuals. The study presents the time series approach with the help of historical prices on the Bombay Stock Exchange's (BSE) Health Care Index, both in the long and short run, using the ARIMA model. The period of the study is from February 1999 to August 2020. The ARIMA equations are used to forecast the future price movement of the Health Care Index till December 2020. The findings reveal that the market will continue with the same volatility, and investors should give due attention to analysis and logical reasoning rather than following their feeling of overconfidence.
Some of the most interesting news from the frontiers of machine intelligence is about the triumphs in the domains traditionally considered "creative." The attempts to generate aesthetically pleasing patterns and images using computers are as old as the invention of the computer itself. With the advent of powerful AI algorithms, it is now possible to train neural networks to generate images not only in a particular artistic style but in completely new styles as well. In other realms of human creativity, algorithms are being used in "proving" mathematical theorems, composing music, writing screenplays, and generating snippets of poetry. This chapter outlines the history of computer-generated art, followed by the recent advances in machine intelligence as applied to the creative fields. Finally, it briefly discusses the philosophical, ethical and legal questions related to ownership of AI-generated or AI-assisted creative work.
A significant role in clinical treatment and educational tasks is played by clinical image classification. However, the traditional approach has reached its peak in terms of implementation. Additionally, using traditional approaches requires a lot of time and effort to remove and choose arrangement features. The deep learning (DL) model is a new machine learning (ML) technique that has proven effective for various classification problems. To alter image classification problems, the convolutional neural network performs well, with the best results. This chapter discusses the importance and challenges of deep learning models in medical image classification and explains some techniques for reducing overfitting and leveraging model performance during model training.
This chapter describes the various ways used in streaming analytics and presents the various real-time streaming analytics that are being used in both healthcare and in the retail industry. Big Data analytics work with both structured and unstructured data sources. Recently, both researchers as well as medical professionals have adopted real-time data streaming analytics and have improved their decision-making to save lives. The main challenge is the diversity of data from medical images, ranging from physiologic and clinical data. In the retail industry, streaming analytics are used for inventory tracking, customer service, warehousing tracking and throughout the whole production process.
Closed-circuit television (CCTV) was installed to inspect the normal performance of sewer pipes by continuously monitoring the type and location of pipe defects, such as cracks, roots, deposits, and infiltrations. Through the digital images produced by this method, the identification of sewer pipe defects is based on deep learning techniques, namely faster region-based convolutional neural network (faster R-CNN), and different CNN architectures have been combined with faster R-CNN to produce better performance. After training, this model is evaluated in terms of detection accuracy, achieving a Mean Average Precision (MAP) of 90 per cent, using the VGG network in the sewer pipe system.
The rapid proliferation of data from applications including IoT, and on-demand access to data have increased dependency on cloud computing, which helps to minimize the overhead related to data storage and maintenance. Applications such as IoT, industrial control, etc. generate data which are highly time-critical in most scenarios. The cloud platform offers permanent storage of this massive amount of data but with comparatively less focus on time-sensitivity. Edge/fog computing are extensions of the cloud computing paradigm and require less response time for time-sensitive data. The edge/fog brings processing and storage closer to the edge of the network, thereby reducing network traffic, delay, and latency. It acts as an intermediate layer between the end devices and the cloud platform, for data collection, offloading, processing, and data management. This chapter addresses the need for fog computing, presents the design model for edge/fog computing, and discusses applications and open issues of implementation. The three-layered network model, the services provided by the edge/fog computing, and a few research challenges of implementation will also be discussed.
Human-computer interaction (HCI) is an area of computer science that investigates how people interact with computers. Augmented reality (AR) and ubiquitous computing (UC) are two of the most fundamental and important fields of HCI. Sensors are always developing, becoming smaller, lighter, more precise, long-lasting, effective, reactive, and with enhanced communication capabilities. The best examples include smart watches, smart rings, advanced medical gadgets, and earphones. UC has evolved and has many applications, such as in healthcare, accessibility, in learning, and in logistics. AR is a system that blends the real and virtual worlds, providing real-time interaction and accurate 3D detection of virtual and real objects. AR has a bright future because of its real-time interaction with the virtual. Businesses globally have now begun developing and selling their products using augmented reality technology.
Self-driving vehicles are a familiar sight nowadays, and they function with the help of Artificial Intelligence. Vehicles with human drivers have caused many accidents, so in that case, self-driving vehicles or driverless vehicles need to be more intelligent and be careful if they meet any obstacle on the road while driving. Detecting objects is necessary for self-driving vehicles. This object detection is a part of the computer vision that plays an important role in finding objects or obstacles. Computer vision with the Deep Learning techniques, based on the object detection model of YOLO is the latest and fastest object detection algorithm. This work has been proposed using YOLO models to detect objects. This chapter discusses the five classes that were taken as an object to be detected in a traffic environment. From the captured video, 200 image frames were received and given as input image frames to YOLO models. The work involves the concept of using two models, YOLOv4 and YOLOv5. Compared to other YOLO models the YOLOv5x model performs well and achieved above 92 per cent of precision. The image frames given to the YOLOv4 network detected the objects correctly and also produce the audio segment of the labelled object's text output.
Abstract Consider a dynamic system, such as a crane or a plane, and an operator who has acquired the skill of operating that system. Suppose then that the operator is asked to design an algorithm for automatically controlling the system. The operator may try to design such an automatic controller by reconstructing his or her subcognitive skill through introspection, or by other means, e.g. by reasoning about the system and engineering the controller. The main question investigated in this paper is, how would different operators go about this task, and how does the success depend on various factors affecting the task? These factors include:
Abstract On l October 1945 Turing was appointed to the newly-formed Mathematics Division of the National Physical Laboratory, his brief to design an electronic stored-program digital computer. The lectures published here, given by Turing and his assistant J.H. Wilkinson in December 1946-February 1947, add substantially to our knowledge of Turing’s design. The lectures detail the evolution of the design from Version V of early 1946 through Version VI to Version VII. On 8 December 1943, the world’s first large-scale special-purpose electronic digital computer came into operation, at the Government Code and Cypher School, Bletchley Park, England.
Abstract World Wide Web (WWW) usage is increasing rapidly. Users waste time downloading pages that turn out to have no interest to them. Interesting pages are often overlooked. Modern browsers allow the user to specify search strings. This paper experimentally investigates a different approach based on learning user WWW pages preferences from examples. WWW users were drawn from a class of students. The experiments show that the inductive logic programming algorithm Progol gives overall significant predictive accuracy of user interests. However, the results are highly polarized. Some users are very predictable and others not. The polarization was surprisingly found to correlate in all cases with student exam performance. This work was conducted as part of Parson’s MSc thesis [7].
Abstract The discovery of an underlying law from a set of numeric data is the central part of scientific discovery systems. This paper proposes a new connectionist approach to numeric law discovery. In order to efficiently and constantly obtain near-optimal results (law candidates), we introduce a new second-order learning algorithm; by adopting a quasi Newton method as a basic framework, the optimal step-lengths are calculated as the minimal points of second-order approximations. The minimum -description length criterion selects the most suitable from law-candidates. The main advantage of our method over previous work on the symbolic or connectionist approach is that it can efficiently discover numeric laws whose power values are not restricted to integers. Experiments showed that the proposed method works well in discovering such laws even from data containing a small amount of noise.
