Opinion mining encompasses a wide variety of subfields, one of which is known as sentiment analysis. In this subfield, the focus of the investigation is on extracting the feelings and beliefs of the general public regarding a specific topic from structured, semi-structured, or unstructured textual information. The database of IMDB movie reviews serves as the primary focus of our sentiment analysis study, which is included in this thesis. In the work that we did for our thesis, we provide a fresh method to an enhanced version of the Naive Bayes algorithm. This is accomplished with the help of Tf-IDF (Term Frequency-Inverse Document Frequency). The comparison is carried out on datasets of special sizes, and it is carried out on the basis of criteria such as mean square error, accuracy, precision, recall, and F1 score. Our work has shown superior accuracy than other classification algorithms, and the comparison has been carried out.

Image restoration resolves the issue of unsuitable scene portrayal. The objective of image restoration is to control an image so that it will in some sense all the more intently portray the scene that it addresses. The image restoration issue shows up in many fields. Practically all disciplines wherein images are gained under not so great circumstances find restoration methods valuable-space science, medication, criminology, and military observation, for instance. Photo printing labs may likewise find restoration procedures a practical apparatus in cleaning up extraordinary photographs.

Some observers are beginning to focus their attention on web-based media as a result of the current circumstances. Massive amounts of information can be generated on the client side of web-based media platforms. In order to stay up with the information that the customers produce, there are a lot of mining errands that are offered through online media mining. There are many different long-distance interpersonal communication destinations where the customer can create their own local region based on what is most convenient for them. As it has come to be known, web-based media constitutes a significant portion of the whole virtual world. This is due to the fact that a large number of users maintain their own profiles and participate in a variety of online communities. The primary characteristic of online media is that of community discovery. It is comparable to the clustering stage of the information mining process. However, identifying local area by effect is another technique of identification that can be used in online media mining. A lot of work has been done on local area recognition based in part on this, but it is still important. The primary objective of the exposition is to honor those in the surrounding area who are putting Leverage to good use. Scalability and the quality of the community are also important concerns of community detection. Some of the algorithms are scalable in big networks and produce superior results than other algorithms when compared to those findings. We have analyzed the differences and similarities between the algorithms based on the social data from Twitter. As a consequence of this, it has been demonstrated that the algorithms are scalable in the big network according to the evaluation parameters. One thing that sets this thesis apart from others is the fact that we have tested each and every aspect of the algorithm on a significant social network.

Recognizing the activities of human beings is viewed as extremely fundamental in person-to- person communication and relational relationships as it delivers information with respect to the individuality of a group, their character and mental condition. Digging out this data is not an easy job. The significant concept of investigation of AI and machine Learning is the capability of a person to distinguishing the exercises of someone else. The human gestures are complex and also very dynamic. This paper discusses thoughts of different researchers.

Youth enlistment in the military after completing high school is an urgent necessity. As a result, the country would maintain a huge reserve force to protect its borders. The spirit of patriotism, along with high moral values, ethics, and honesty, as well as strong discipline, will ensure that our kids are productive, utilise their time wisely, and contribute to the self-sufficiency of our nation. As a result, military conscription should be made mandatory for all graduates of Class 12.

In this paper a single phase active power filter is connected to a single phase grid with non-linear load. The single phase active power filter mitigates the harmonics generated by the non-linear load connected at PCC. The harmonics generated by the non-linear load are redirected to active power filter direction so as to avoid injection into the source damaging it. The novel single phase active power filter has half bridge connected to share capacitors. The two power electronic devices of the active power filter are controlled by parabolic PWM with feedback from the source voltage and load current for compensation of harmonics in source current. The conventional PI controller is updated with fuzzy logic controller for further reduction of harmonics. The circuit of both the controllers with active power filter connected to single phase grid is modelled in MATLAB Simulink environment with graphs generated with respect to time.

In this paper a single-phase transformer less PVA integrated inverter is connected to the grid with grid voltage synchronization using PLL. The topological structure of the inverter ensures that the common mode voltage does not contain high frequency components, thereby reducing the magnitude of leakage current involved with the solar panels well within the acceptable limit. The PI controller is further replaced with fuzzy logic controller for better improvement in power injection. A Simulation analysis is carried out on the converter using MATLAB software.

The main cause of death for women is breast cancer. When this condition is identified early with the aid of mammography, the death rate is decreased. As the number of patients rises, radiologists find it more challenging to complete the diagnostic procedure in the limited time allowed. This study's objective is to investigate the various neural network based techniques that can help radiologists diagnose breast cancer more quickly and accurately (DL). This study compares multiple machine learning and deep learning methods in order to determine the most effective classifier for breast cancer categorization.