Label Hallucination for Few-Shot Classification
Authors: Yiren Jian, Lorenzo Torresani7005-7014
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Despite the simplicity of the approach, we show that that our method outperforms the state-of-the-art on four well-established few-shot classification benchmarks. The code and appendix are available at https://github.com/yiren-jian/Label Halluc. |
| Researcher Affiliation | Academia | Yiren Jian, Lorenzo Torresani Dartmouth College yiren.jian.gr@dartmouth.edu, LT@dartmouth.edu |
| Pseudocode | No | The paper includes Figure 1 as an 'Overview of our proposed approach' which illustrates steps, but it is a diagram, not structured pseudocode or an algorithm block. |
| Open Source Code | Yes | The code and appendix are available at https://github.com/yiren-jian/Label Halluc. |
| Open Datasets | Yes | We evaluate our method on four widely used few-shot recognition benchmarks: mini Image Net (Vinyals et al. 2016), tiered Image Net (Ren et al. 2018), CIFAR-FS (Bertinetto et al. 2019), and FC100 (Oreshkin, L opez, and Lacoste 2018). |
| Dataset Splits | No | The paper defines support and query sets within episodes as training/testing, but it does not specify the overall training/validation/test splits (e.g., percentages or exact counts) for the datasets used (mini ImageNet, CIFAR-FS, etc.). While tables mention 'Training is done on the training split only', this does not provide specific split information for reproduction. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for experiments, such as GPU/CPU models or cloud instance types. |
| Software Dependencies | No | The paper does not provide specific software dependencies or their version numbers, beyond mentioning the use of Res Net-12. |
| Experiment Setup | Yes | Finetuning the Whole Model to Recognize Novel Classes... where ˆy denotes the hallucinated pseudo-label, LKL is the KL divergence between the predictions of the model and the pseudo-labels scaled by temperature T, and α, β are hyperparameters trading off the importance of the two losses. Since the support set is quite small... we use data augmentation to generate multiple views of each support image, so as to obtain enough examples to fill half of the mini-batch. Specifically, we adopt the standard settings used in prior works... and apply random cropping, color jittering and random flipping to generate multiple views. |