Search Results for author:
Found 5 papers, 0 papers with code
Discerning and Resolving Knowledge Conflicts through Adaptive Decoding with Contextual Information-Entropy Constraint
Large language models internalize enormous parametric knowledge during pre-training.
Learning Representation for Anomaly Detection of Vehicle Trajectories
We conduct extensive experiments to demonstrate that our supervised method based on contrastive learning and unsupervised method based on reconstruction with semantic latent space can significantly improve the performance of anomalous trajectory detection in their corresponding settings over various baseline methods.
Pay attention to emoji: Feature Fusion Network with EmoGraph2vec Model for Sentiment Analysis
Based on EmoGraph2vec model, we design a novel neural network to incorporate text and emoji information into sentiment analysis, which uses a hybrid-attention module combined with TextCNN-based classifier to improve performance.
Emoji-based Co-attention Network for Microblog Sentiment Analysis
In this paper, we propose an emoji-based co-attention network that learns the mutual emotional semantics between text and emojis on microblogs.
Face De-occlusion using 3D Morphable Model and Generative Adversarial Network
In recent decades, 3D morphable model (3DMM) has been commonly used in image-based photorealistic 3D face reconstruction.
