Deep Quantization Network for Efficient Image Retrieval
Authors: Yue Cao, Mingsheng Long, Jianmin Wang, Han Zhu, Qingfu Wen
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on standard image retrieval datasets show the proposed DQN model yields substantial boosts over latest state-of-the-art hashing methods. |
| Researcher Affiliation | Academia | School of Software, Tsinghua University, Beijing, China Tsinghua National Laboratory for Information Science and Technology |
| Pseudocode | No | The paper contains mathematical formulations and derivations but no structured pseudocode or algorithm blocks. |
| Open Source Code | No | The codes and configurations will be made available online. |
| Open Datasets | Yes | NUS-WIDE1 is a public web image dataset. CIFAR-102 is a dataset containing 60,000 color images in 10 classes... Flickr3 consists of 25,000 images collected from Flickr... |
| Dataset Splits | No | In NUS-WIDE and CIFAR-10, we randomly select 100 images per class as the test query set, and 500 images per class as the training set. In Flickr, we randomly select 1000 images as the test query set, and 4000 images as the training set. The paper specifies training and testing sets, but does not explicitly mention a separate validation set with specific counts or percentages for reproduction. |
| Hardware Specification | No | The paper does not explicitly describe the specific hardware (e.g., GPU/CPU models, memory) used to run the experiments. |
| Software Dependencies | No | We implement the DQN model based on the open-source Caffe framework (Jia et al. 2014). The paper mentions using the Caffe framework and Alex Net architecture but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | Yes | As the fch layer is trained from scratch, we set its learning rate to be 10 times that of the lower layers. We use the mini-batch stochastic gradient descent (SGD) with 0.9 momentum and the learning rate annealing strategy implemented in Caffe, and cross-validate learning rate from 10−5 to 10−2 with a multiplicative step-size 10. We also fix the mini-batch size of images as 64 and the weight decay parameter as 0.0005. For the product quantization loss, we cross-validate the λ from 10−5 to 1 with a multiplicative step-size 10. |