Prototypical Examples in Deep Learning: Metrics, Characteristics, and Utility
By Nicholas Carlini, Ú. Erlingsson, Nicolas Papernot. 2018
25 citations recorded by the source corpus.
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Nicholas Carlini
- Florian Tramèr
- Yoshua Bengio
- Nicolas Papernot
- Bernt Schiele
- Bo Li
- Quoc V. Le
- Percy Liang
Citations in the corpus
- Adversarial Training and Robustness for Multiple Perturbations (2019, background): “This hints that some points (sometimes referred to as prototypical examples [4, 36]) are inherently easier to classify robustly, regardless of the perturbation type.”
- Does learning require memorization? a short tale about a long tail (2019, background): “A thorough and recent exploration related to this question can be found in thework of Carlini et al.[16].”
- Characterising Bias in Compressed Models (2020, background): “Leveraging a subset of data points to understand model behaviour or to audit a dataset fits into a broader literature that aims to characterize input data points as prototypes – “most typical" examples of a class – (Carlini et al., 2019; Agarwal & Hooker, 2020; Stock & Cisse, 2017; Jiang et al., 202”
- High Accuracy and High Fidelity Extraction of Neural Networks (2019, background): “[44], we notice that test points where fidelity is easiest to achieve are also the most prototypical (i.”
- High-Fidelity Extraction of Neural Network Models (2019, methodology): “Using prototypicality metrics, as investigated in Carlini et al. [43], we notice that those points where fidelity is easiest to achieve are also the most prototypical (i.e., more representative of the class it is labeled as).”
- Selective Brain Damage: Measuring the Disparate Impact of Model Pruning (2019, background): “…fits into a broader and non-overlapping body of literature that aims to classify input data points as prototypes – “most typical” examples of a class – ([3, 63]) or outside of the training distribution (OOD) [30, 46, 49, 46, 52] and work on calibrating deep neural network predictions [45, 20, 40].”