Emoticon Prediction on Textual Data Using Stacked LSTM Model

Mittal, Mamta; Arora, Maanak; Pandey, Tushar

doi:10.1007/978-981-15-3325-9_20

Cited by 3 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, the paper cites a study which used an SVM (Support Vector Machine) model to achieve an accuracy of 82.3% and another which used a BLSTM-CNN model to achieve an accuracy of 87.8%. The author of this paper [12] describes a method for predicting emoticons on textual data using a stacked Long Short-Term Memory (LSTM) model. The stacked LSTM model is composed of multiple LSTM layers, which are used to process the text and generate an output corresponding to a particular emoticon.…”

Section: Related Workmentioning

confidence: 99%

Emoji Prediction Using Bi-Directional LSTM

Kone,

Anagal,

Anegundi

et al. 2023

ITM Web Conf.

View full text Add to dashboard Cite

Messengers and social media dominate today’s internet usage across the globe. For the large population, a typical day starts with messages flooding on mobiles, from simple good morning wishes, business meeting invites, reminders, and schedules for the day and the list is endless. A striking feature of today’s digital communication is the variety of emojis used, without which text communication almost look incomplete. Emojis are graphic symbols/logograms used with text communication to enhance the effectiveness of emotions and set an undertone that makes texting a more fun experience for the users. Emojis are the visual language of the new generation. They give consumers a means to communicate their feelings while reducing the quantity of text that needs to be typed by the sender. Every social media and messenger platform like Facebook, Instagram, Twitter, WhatsApp, and many more have its own emoji set. To lure more and more users, many new emojis are added day by day. Predicting and suggesting emojis based on the text, emotion and user patterns to the user is an important feature of today’s messengers and social media applications. If you start typing a message, relevant emojis will be displayed from which users can choose an emoji, further enhancing the user texting experience. This process is done using natural language processing and machine learning techniques. In this paper, we study emoji prediction techniques and propose an emoji prediction model using bi-directional LSTMs. We compare emoji prediction NLP techniques, including RNN, LSTM, LSTM networks, and Bi-LSTM. Based on our implementation, we suggest that the bi-directional LSTM model is the most effective technique. Our model outperforms many baseline approaches with an accuracy of 94% when tested on a CodaLab Twitter data set with 60000 rows and two columns. Our study shows the effectiveness and efficiency of bi-directional LSTMs for text-based systems for communication.

show abstract

Section: Related Workmentioning

confidence: 99%

Emoji Prediction Using Bi-Directional LSTM

Kone,

Anagal,

Anegundi

et al. 2023

ITM Web Conf.

View full text Add to dashboard Cite

show abstract

“…The principle behind the LSTM architecture is a memory cell which can sustain its state over time [40], and non-linear gating units which manage the information flux into and out of the cell. A deep network of convolutional LSTM can be used to access the full spectrum [41] of temporal information at spatial scales of the data.…”

Section: Detailed Architecturementioning

confidence: 99%

DeepFake Video Detection: A Time-Distributed Approach

et al. 2020

Self Cite

View full text Add to dashboard Cite

Recent developments in machine learning algorithms have led to the generation of forged videos having remarkable quality, which are indistinguishable from real videos. This can fatally affect the way in which one perceives the information available digitally. Thus, this paper aims to efficiently and holistically detect manipulated videos generated using DeepFake, which is the most effective deep learning powered technique developed so far by the researchers. Arduous efforts have been put to detect the forgery in still images, but the authors leveraged the spatio-temporal features of the videos by taking sequences of frames as input to the model. Furthermore, the authors have proposed an architecture which took advantage of lower-level features around regions of interest as well as discrepancies across multiple frames. Experiments have been performed on the Deep Fake Detection Challenge dataset of ≈ 470 GB in size, and it has been observed that the proposed approach yielded a test accuracy score of 97.6%.

show abstract