IEEE

ACIR-MACL: Effective Multimodal Sentiment Analysis via Attention-Based Causal Intervention Regularization and Multi-Aspect Contrastive Learning

Abstract: Multimodal Sentiment Analysis (MSA), which involves interpreting sentiment from language, vision, and acoustics, is a challenging task. While existing models have achieved considerable ...

Small Language Models Outperform Frontier AI On Cost, Speed And Accuracy

Bigger has defined AI from day one. New data says task-specific small models beat frontier LLMs on accuracy, cost and speed — and save money.
GitHub

Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search

This repository releases the official code for Mini-o3. We achieve the state-of-the-art results on various benchmarks and present a full training recipe to reproduce the OpenAI o3-style deep ...
IEEE

Twitter Sentiment Analysis using Machine Learning Techniques

Abstract: Sentiment analysis involves identifying and classifying opinions or sentiments expressed in textual data. With the rise of social media platforms, vast amounts of sentiment-rich data are ...