Название: Advances in Data Science and Artificial Intelligence
Автор: Rajiv Misra, Nishtha Kesswani, Muttukrishnan Rajarajan, Bharadwaj Veeravalli
Издательство: Springer
Год: 2023
Страниц: 521
Язык: английский
Формат: pdf (true)
Размер: 15.0 MB

With the intriguing development of technologies in several industries along with the advent of accrescent and ubiquitous computational resources, it creates an ample number of opportunities to develop innovative intelligence technologies in order to solve the wide range of uncertainties, imprecision, and vagueness issues in various real-life problems. Hybridizing modern Computational Intelligence with traditional computing methods has attracted researchers and academicians to focus on developing innovative AI techniques using Data Science. International Conference on Data Science and Artificial Intelligence (ICDSAI) 2022, organized on April 23-24, 2022 by the Indian Institute of Technology, Patna at NITIE Mumbai (India) in collaboration with the International Association of Academicians (IAASSE) USA collected scientific and technical contributions with respect to models, tools, technologies, and applications in the field of modern Artificial Intelligence and Data Science, covering the entire range of concepts from theory to practice, including case studies, works-in-progress, and conceptual explorations. The main conference included over four technical sections that focused on fundamentals of Data Science applications in mechanical engineering, ML and Artificial Intelligence (AI), BOT, Web development, app development.

There are many methods available now to analyze tabular data. The neural networks are highly efficient and provide good accuracy. In this study, artificial neural network (ANN) is chosen for analysis. As we know artificial neural network models are efficient with big sized datasets, but they face problems when dealing with smaller datasets. The problem faced with some of the smaller datasets is imbalance as well. This study focuses on achieving the solution for the same. The below study with various sections gives us a chance to grasp the results of 13 standard datasets which are imbalanced in nature. The next section, Literature Survey, throws light on various ways in which artificial neural network models can solve the problem of imbalance in datasets. The following section, Methodology, talks about artificial neural network, TabNet, and two other customized neural networks.

S.A.R.A (Smart AI Refrigerator Assistant): The fast pace of life has increased the popularity of instant foods. The recent quarantine period has piqued the interest of many in cooking. When it comes to food, deciding what recipe to make often takes longer than the actual preparation time. To save this time and present a wide variety of options to the user, S.A.R.A, an AI-powered refrigerator assistant, has been developed. Its features encompass “Recipe Recommendation” based on the user input. This assistant can interact with the user through a progressive web application. The data was processed remotely on a server. The data required for the recommendation algorithm was acquired by leveraging the web scraping technologies using open-source Python libraries. This data was then stored in JSON files. The textual data was organized with data wrangling and analysis techniques. The recipe recommendation functionality was achieved by developing a search engine specific to food recipes through extensive Natural Language Processing (NLP). Further, the technologies used were completely open source, and the modifications which will be required to integrate our assistant are minimal as a result of which the whole idea is extremely budget-friendly. Introducing such a feature will not just save time but also entice the user to try newer recipes. The user can explore and include these newly discovered recipes in his or her diet. The goal is not only restricted to comfort and ease but also aims toward the accomplishment of a healthier lifestyle. In addition to this, a variety of additional features can be introduced in this assistant which will ensure even more ease and benefit for the user. This paper provides the detailed approach, methodology, system architecture, and an in-depth analysis of the functionality this assistant is capable of.