Extensible Cross-Modal Hashing

Authors: Tian-yi Chen, Lan Zhang, Shi-cong Zhang, Zi-long Li, Bai-chuan Huang

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our extensive experiments show the effectiveness of our design. and sections like 3 Experiment
Researcher Affiliation Academia 1School of Computer Science and Technology, University of Science and Technology of China, China 2School of Data Science, University of Science and Technology of China, China 3School of Information Science and Engineering, Northeastern University, China 4Department of Physics, University of California Berkeley, USA
Pseudocode Yes Algorithm 1 Core Algorithm and Algorithm 2 Extending Model for New Tasks.
Open Source Code No Not found. The paper does not provide an explicit statement or link to open-source code for the described methodology.
Open Datasets Yes The MIRFLICKR-25k dataset [Huiskes and Lew, 2008]... We also adopt the MSCOCO-2014 dataset [Lin et al., 2014]
Dataset Splits Yes In experiments using MIRFLICKR-25k, 10,015 instances are randomly chosen as the train set, and the rest 10000 are used for validation, namely 2000 for the query and 8000 for the database. In experiments using MSCOCO, 16,869 randomly chosen instances are used for training, and the rest 5000 and 15000 instances are used as query and database, respectively.
Hardware Specification Yes All experiments are conducted on a server with 4 TITAN X GPUs.
Software Dependencies No We implement ECMH via Pytorch.
Experiment Setup Yes We set all learning rate to 1.5 and decrease it by 5% every 100 steps. α is set to the range of [0.1, 0.15] and β is set to 0.5. Batch size is fixed to 500.