Deep-HOSeq: Deep Higher-Order Sequence Fusion for Multimodal Sentiment Analysis --- ICDM 2020
Deep-HOSeq is written in python3 with some code fragments copied from DCGAN implementation from Carpedm20.
- The code is developed with Python 3.6 and TensorFlow 1.12.0 (with GPU support) on Linux
- For reasons of my convenience,
data_diris required to bedata_dir = ../../data-- errors might pop-up when other directories are used. - To train the model please execute main_DeepHOSeq.py
Please download the dataset from CMU-MultimodalSDK https://github.com/A2Zadeh/CMU-MultimodalSDK
- Better Fusion scheme for utilizing both common and unique latent information
- Utilize Sequence information for sentiment prediction
- Cross Data Generelaization Performance
Please contact either Sunny Verma or Wei Liu at firstname.lastname@uts.edu.au if you're interested to collaborate on this!