• Mrs. A. Sureka selas
• Mrs. S. Suguna
• Ms. T. Sowmia
• Ms. R. Srilekha

Forms of digital media such as blogs, online news media, and social media have taken the place of former news delivery platforms such as newspapers and magazines. The linked recommendations of these platforms help spread the word in no time. This fascination leads to an invasion of negativity such as fake news and misrepresented information. False information negatively affects all areas such as health care, education, government, and the market because people make decisions about anything based on available information. Messages can be text or multimodal. Any combination of text, image, and video can be present in multimodal messages. Attention seekers create fake news by altering text, images, or both. The spread of fake news is in different domains, but fact-checking sites can verify the authenticity of a particular environment; thus, detecting fake news remains challenging. Another reason for the difficulty in detecting fake news is the unstructured representation of news (in the form of articles, images, audio, video, etc.) that a person needs to classify. Despite the large amount of research work that has been done to meet this purpose, proper classification still faces various challenges such as imbalance, multimodality, lack of appropriate structure, and ambiguity of words in datasets. This research proposes four novel deep learning architectures for text and image-based fake news classification. Headlines explain news headlines and draw anyone's attention to some information. The image provides relevant image data about the news. In many fake news, the textual content and visual information will not be related. This work is a hybrid model that uses all three pieces of information, namely title, text, and image. In the first work, a Deep-Learned Bidirectional Gated Recurrent Unit (Bi GRU) - Long ShortTerm Memory (RNN) Model (DL-BGLM) is used to detect fake messages using textual content and message title. Detection is done by incorporating two new subframes. The first sub-framework of the Glove Embedded Bidirectional Predicted Attention Scheme (GEBPA) gets significant information from the name. The second subframe Multi-Layered Convolution and Bidirectional LSTM (MCBL) scheme extracts salient features from the text.

Fake News Detection, Deep Features, Ensemble based methods, Machine Learning, RNN

"MACHINE LEARNING BASED FAKE DETECTION MODEL USING REGIONAL CONVOLUTION NEURAL NETWORK ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 5, page no.h154-h165, May-2024, Available :http://www.jetir.org/papers/JETIR2405719.pdf

"MACHINE LEARNING BASED FAKE DETECTION MODEL USING REGIONAL CONVOLUTION NEURAL NETWORK ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 5, page no. pph154-h165, May-2024, Available at : http://www.jetir.org/papers/JETIR2405719.pdf