Fine-grained Image Classification by Visual-Semantic Embedding

Authors: Huapeng Xu, Guilin Qi, Jingjing Li, Meng Wang, Kang Xu, Huan Gao

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on a challenging large-scale UCSD Bird-200-2011 dataset verify that our approach outperforms several stateof-the-art methods with significant advances.
Researcher Affiliation Academia 1 Southeast University, Nanjing, China 2 University of Electronic Science and Technology of China, Chendu, China 3 Xi an Jiaotong University, Xi an, China 4 Nanjing University of Posts and Telecommunications, Nanjing, China
Pseudocode No The paper does not contain any clearly labeled pseudocode or algorithm blocks. Procedures are described in narrative text and mathematical equations.
Open Source Code No The paper does not provide any statement about releasing source code or a link to a code repository.
Open Datasets Yes We choose DBpedia [Lehmann et al., 2015] (KB) and English-language Wikipedia (text) from 06.01.2016 as external knowledge. Word2Vec and Trans R (described in Section 4) are used to get the class embedding. In this section, we present the experimental settings and show experimental results of our proposed model on the widely-used benchmark Caltech-UCSD Bird-200-2011 [Wah et al., 2011].
Dataset Splits No The paper mentions using "Caltech-UCSD Bird-200-2011" and training, but does not specify the exact training/validation/test splits (e.g., percentages or sample counts) used for reproduction.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU model, CPU type) used for running the experiments.
Software Dependencies No The paper mentions several deep learning architectures and techniques (e.g., Alex Net, VGG, Google Net, Res Net, Word2Vec, Trans R, batch-normalization, dropout), but does not provide specific version numbers for any underlying software dependencies (e.g., Python, TensorFlow, PyTorch).
Experiment Setup Yes We train our model using stochastic gradient descent with mini-batches 40 and learning rate 0.0015. The hyperparameter α of Eq. 7 is set to be 0.85 with cross-validation.