Multi-Modal Knowledge Hypergraph for Diverse Image Retrieval

Authors: Yawen Zeng, Qin Jin, Tengfei Bao, Wenfeng Li

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on two real-world datasets have well verified the effectiveness and explainability of our proposed method.
Researcher Affiliation Collaboration 1 Byte Dance AI Lab 2 School of Information, Renmin University of China
Pseudocode No The paper describes its methods and components in text and mathematical formulas but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper states 'We implement our solution based on the Tensorflow framework3.', but the footnote 3 links to the Tensorflow framework itself, not to the authors' specific implementation code for their proposed method.
Open Datasets Yes Div150Ad Hoc1 is a dataset for the competition of diverse social image retrieval (Ionescu et al. 2016), which contains a variety of keyword-based queries... 1http://campus.pub.ro/lab7/bionescu/Div150Adhoc.html. Div4002 is constructed by Media Eval Workshop (Ionescu et al. 2014)... 2http://multimediaeval.org/mediaeval2014/diverseimages2014
Dataset Splits No The paper mentions using 'training batches' for losses and specifies hyperparameters used on the datasets, but it does not explicitly provide the train/validation/test dataset splits (e.g., percentages or sample counts) used for reproducibility.
Hardware Specification No The paper does not provide any specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments.
Software Dependencies No The paper mentions 'Tensorflow framework' but does not specify a version number for Tensorflow or any other key software dependencies with their versions.
Experiment Setup Yes For specific hyperparameters in our method, the instance numbers m, the margin , the threshold τ and the balance factors λ1, λ2 are set as (8, 0.4, 0.4, 0.4, 0.3) and (8, 0.4, 0.3, 0.3, 0.3) on Div150Ad Hoc and Div400 datasets, respectively.