Contents I. Introduction
The evolution of technology should not only give the nod to the progress that was made in the past but also raise concern about their impact on the confidentiality and protection of this data. While we will start with Federated Learning (FL) and the well-known FedAvg algorithm, we need to keep in mind the elements that prove the difficulty in understanding these components and how imperative such analysis can be for the reader [1]. In today's technological age, data generation and implementation have progressed tremendously and machine learning is at the crux of this transformation. Machine intelligence that underpins most of the state-of-the-art technologies, is empowered by billions of data points, comprising of recommendation systems, intelligent robots and self-driving cars [2]. Despite the reliance on collecting and analyzing data, the fact that this can give rise to many privacy and security issues is something worth noting [3]. The analogue centralized data management approach of data collection from numerous sources for analysis opens opportunities to make connections, data tampering, and inappropriate data usage [4]. Importance: Privacy and security issues are some of the factors that have made people hesitate about participating in 5G networks. It demonstrates that the conflict between privacy and safety is one of the main reasons to make machine learning sustainable. In today's security-conscious world, the concern of consumers has gradually risen, leading to the superior safety of data overall. It affects not only rights like privacy but may infringe other rights like intellectual property and the safety of national secrets. A passage that the potential of AI to offer a way forward is highlighted and such innovations should not undermine the private rights and security of the society. Federated Learning: Change the Game to a Big Paradigm [4]. With a little change, Federated Learning is now becoming the latest theory applied in the context of local data technologies. Florida is a basic conceptual shift from the vertical traditional authorities/centralized approaches where data is processed at the highest point of the urban area to the horizontal one. Federated Learning (FL) takes on these difficult privacy and security questions in a direct manner. As model training is carried out on numerous devices or servers without the need for communication of real data, the necessity of communicational exchange of actual data is eliminated [5]. Such a method makes it possible to store data locally on devices, while communication concerns transmit modifications to the cloud for overall aggregation purposes. This technique also prevents data breaches to some extent by providing an open way to use the data that was previously highly restricted due to confidentiality reasons [6]. The principle of Federated Averaging (FedAvg) is the central building block of the Federated Learning (FL) structure [7]. FedAvg algorithm which serves a key function in the network called Federated Learning system, is acknowledged as the most important constituent within the system. In such a way it serves as an [8] FL infrastructure which allows to use of model updates from separate sources which eventually leads to the improvement of a global model [9]. The implementation of this strategy makes sure that the total learning value from the cumulative data is retained while the complete individual points remain hidden [10]. The guest lecturer will define the way the FedAvg algorithm proceeds not merely as a technical approach but also as an idea to achieve secure machine learning. The main gameplay of employing FedAvg to object labelling is based upon showcasing the ability to perform two diverse tasks at a time. Initially, the paper serves as a consolidated platform for discussing Federated Learning from a practical and easily intuitive area i.e. machine learning, in a manner that applies to all. There are many cases of image classification where the applications, among others, include filtering social media content and diagnosing medical conditions [11]. This brings a bright light that illustrates the benefits that are provided by FL. Additionally, this pick says in a loud way that FL became a game changer in several areas surrounding the need for data privacy using ML in places where it used to be impossible before. Thus, It expressed the fact that distributed Averaging (FedAvg) [12] is theoretically possible and can have a considerable effect. The introduction concludes by articulating the aim of the paper: to probe into the exploitation and efficiency of the FedAvg algorithm in the classification of pictures, let us however, focus on the importance of such an algorithm in solving intricate data privacy and security issues [13]. This sets up a context for the reader to do more than just grasp the contribution of the paper to the overall discussion on privacy-preserving machine learning. In this paper, not only is the FedAvg algorithm subject to a detailed technical description but also its wide-scope implications are explored for the sensitive processes of data privacy, security, and machine learning. The about to present a paper on "Image Classification Using Federated Averaging Algorithm" is successful in terms of obtaining several vital objectives [14]. The article assembles the research in the frame of the difficulties occurring in digital space, reflects upon privacy and security, presents FL and FedAvg as the future of artificial intelligence, illustrates the solution's application to visual recognition, and explains the vision and contributions of the work to the community. On the other hand, such a comparative concept is not only informative but also influential as it impresses on the reader the connection between technological and social progress [15].
