Anonymizing k Facial Attributes via Adversarial Perturbations
Authors: Saheb Chhabra, Richa Singh, Mayank Vatsa, Gaurav Gupta
IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on three popular databases i.e. MUCT, LFWcrop, and Celeb A show that the proposed algorithm not only anonymizes k-attributes, but also preserves image quality and identity information.The proposed algorithm is evaluated on three datasets: MUCT [Milborrow et al., 2010], LFWcrop [Huang et al., 2007], and Celab A[Liu et al., 2015]. As shown in Table 2, three experiments are performed, one corresponding to each case discussed in Section 2. Implementation details: The proposed algorithm is implemented in Tensorflow with 1080 Ti GPU. For learning the perturbation, L2 attack has been performed with Adam optimizer. The learning rate is set 0.01 and number of iterations used is 10000. 4 Performance Evaluation |
| Researcher Affiliation | Collaboration | Saheb Chhabra1, Richa Singh1, Mayank Vatsa1 and Gaurav Gupta2 1 IIIT Delhi, New Delhi, India 2 Ministry of Electronics and Information Technology, New Delhi, India |
| Pseudocode | No | The paper describes the proposed algorithm mathematically and textually through equations and descriptive paragraphs, but it does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any specific link or statement regarding the availability of its source code. |
| Open Datasets | Yes | The proposed algorithm is evaluated on three datasets: MUCT [Milborrow et al., 2010], LFWcrop [Huang et al., 2007], and Celab A[Liu et al., 2015]. |
| Dataset Splits | No | The paper mentions using specific parts of datasets like 'view 2' of LFWcrop and the 'test set' of Celeb A, but it does not provide specific details on training/validation/test splits (e.g., percentages, counts, or explicit methodology for data partitioning) for its own model training. |
| Hardware Specification | Yes | Implementation details: The proposed algorithm is implemented in Tensorflow with 1080 Ti GPU. |
| Software Dependencies | No | The paper mentions that the algorithm is 'implemented in Tensorflow' but does not specify the version number of Tensorflow or any other software dependencies. |
| Experiment Setup | Yes | Implementation details: The proposed algorithm is implemented in Tensorflow with 1080 Ti GPU. For learning the perturbation, L2 attack has been performed with Adam optimizer. The learning rate is set 0.01 and number of iterations used is 10000. |