Frequency Shuffling and Enhancement for Open Set Recognition

Authors: Lijun Liu, Rui Wang, Yuan Wang, Lihua Jing, Chuan Wang

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on various benchmarks demonstrate that the proposed Fre SH consistently trumps the stateof-the-arts by a considerable margin.
Researcher Affiliation Academia 1Institute of Information Engineering, CAS, Beijing, China 2School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China 3Department of Electronic Engineering,Tsinghua University {liulijun, wangrui, jinglihua, wangchuan}@iie.ac.cn, wy23@mails.tsinghua.edu.cn
Pseudocode No The paper describes the proposed methods using detailed text and figures (e.g., Figure 3 for Fre SH overview, Figure 4 for DWT), but it does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statement or link indicating the availability of open-source code for the described methodology.
Open Datasets Yes we evaluate the proposed method on MNIST (Le Cun et al. 2010), SVHN (Netzer et al. 2011), CIFAR10 (Krizhevsky, Hinton et al. 2009), CIFAR+10, CIFAR+50, Tiny-Image Net (Le and Yang 2015).
Dataset Splits Yes Following standard OSR protocols (Guo et al. 2021; Neal et al. 2018), we follow the commonly used splits (Neal et al. 2018).
Hardware Specification Yes All experiments are conducted with NVIDIA RTX 3090 GPU support.
Software Dependencies No The paper mentions using the Adam optimizer and VGG32 as the backbone network, but it does not specify versions for any programming languages, libraries, or other software dependencies.
Experiment Setup Yes We use the Adam optimizer with a batch size of 128 for 600 epochs. The learning rate starts at 0.1 and decays by a factor of 0.1 every 120 epochs.