SmoothGrad: removing noise by adding noise

Smilkov, Daniel; Thorat, Nikhil; Kim, Been; Viégas, Fernanda B.; Wattenberg, Martin

doi:10.48550/arxiv.1706.03825

Cited by 502 publications

(781 citation statements)

References 16 publications

Supporting

Mentioning

778

Contrasting

Unclassified

Order By: Relevance

“…Deterministic visualization methods: Many early works (Erhan et al, 2009;Zeiler & Fergus, 2014;Simonyan et al, 2013;Selvaraju et al, 2016;Smilkov et al, 2017) This led to successful interpretability of DNs internal features, especially when applied on a unit belonging to the first few layers of a DN (Cadena et al, 2018).…”

Section: Related Workmentioning

confidence: 99%

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

Bordes¹,

Balestriero²,

Vincent³

2021

Preprint

View full text Add to dashboard Cite

Discovering what is learned by neural networks remains a challenge. In selfsupervised learning, classification is the most common task used to evaluate how good a representation is. However, relying only on such downstream task can limit our understanding of how much information is retained in the representation of a given input. In this work, we showcase the use of a conditional diffusion based generative model (RCDM) to visualize representations learned with self-supervised models. We further demonstrate how this model's generation quality is on par with state-of-the-art generative models while being faithful to the representation used as conditioning. By using this new tool to analyze self-supervised models, we can show visually that i) SSL (backbone) representation are not really invariant to many data augmentation they were trained on. ii) SSL projector embedding appear too invariant for tasks like classifications. iii) SSL representations are more robust to small adversarial perturbation of their inputs iv) there is an inherent structure learned with SSL model that can be used for image manipulation.Earth from . . . space 1 an untrained representation a supervised representation a SSL representation

show abstract

Section: Related Workmentioning

confidence: 99%

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

Bordes¹,

Balestriero²,

Vincent³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…(3) Gaussian baseline [28,31], (4) uniform baseline [31], and (5) a trained baseline [15], resulting in 15 tables of explanations for each experiment.…”

Section: Comparing Local Explanation Methodsmentioning

confidence: 99%

Topological Representations of Local Explanations

Xenopoulos¹,

Chan²,

Doraiswamy³

et al. 2022

Preprint

View full text Add to dashboard Cite

Local explainability methods -those which seek to generate an explanation for each prediction -are becoming increasingly prevalent due to the need for practitioners to rationalize their model outputs. However, comparing local explainability methods is difficult since they each generate outputs in various scales and dimensions. Furthermore, due to the stochastic nature of some explainability methods, it is possible for different runs of a method to produce contradictory explanations for a given observation. In this paper, we propose a topology-based framework to extract a simplified representation from a set of local explanations. We do so by first modeling the relationship between the explanation space and the model predictions as a scalar function. Then, we compute the topological skeleton of this function. This topological skeleton acts as a signature for such functions, which we use to compare different explanation methods. We demonstrate that our framework can not only reliably identify differences between explainability techniques but also provides stable representations. Then, we show how our framework can be used to identify appropriate parameters for local explainability methods. Our framework is simple, does not require complex optimizations, and can be broadly applied to most local explanation methods. We believe the practicality and versatility of our approach will help promote topology-based approaches as a tool for understanding and comparing explanation methods. CCS CONCEPTS• Mathematics of computing → Topology; • Computing methodologies → Machine learning.

show abstract

“…4. It can be seen that h(x) > h(x) and h(G(z)) < h(G(z)), indicating that the teacher discriminator D is more confident than the student discriminator D. Thus, the teacher discriminator D may More specifically, to illuminate the impact of D and D on G, we visualize the gradients ∇ G(z) h(G(z)) and ∇ G(z) h(G(z)) via SmoothGrad [58] in objective functions Adv(G, D) and Adv(G, D) during the optimization process of Adv(G, D), respectively, shown in Fig. 5.…”

Section: Motivationmentioning

confidence: 99%

DGL-GAN: Discriminator Guided Learning for GAN Compression

Tian¹,

Shen²,

Tao³

et al. 2021

Preprint

View full text Add to dashboard Cite

Generative Adversarial Networks (GANs) with high computation costs, e.g., BigGAN and StyleGAN2, have achieved remarkable results in synthesizing high resolution and diverse images with high fidelity from random noises. Reducing the computation cost of GANs while keeping generating photo-realistic images is an urgent and challenging field for their broad applications on computational resource-limited devices. In this work, we propose a novel yet simple Discriminator Guided Learning approach for compressing vanilla GAN, dubbed DGL-GAN. Motivated by the phenomenon that the teacher discriminator may contain some meaningful information, we transfer the knowledge merely from the teacher discriminator via the adversarial function. We show DGL-GAN is valid since empirically, learning from the teacher discriminator could facilitate the performance of student GANs, verified by extensive experimental findings. Furthermore, we propose a two-stage training strategy for training DGL-GAN, which can largely stabilize its training process and achieve superior performance when we apply DGL-GAN to compress the two most representative large-scale vanilla GANs, i.e., StyleGAN2 and BigGAN. Experiments show that DGL-GAN achieves state-of-the-art (SOTA) results on both StyleGAN2 (FID 2.92 on FFHQ with nearly 1/3 parameters of StyleGAN2) and BigGAN (IS 93.29 and FID 9.92 on ImageNet with nearly 1/4 parameters of BigGAN) and also outperforms several existing vanilla GAN compression techniques. Moreover, DGL-GAN is also effective in boosting the performance of original uncompressed GANs, original uncompressed StyleGAN2 boosted with DGL-GAN achieves FID 2.65 on FFHQ, which achieves a new state-of-the-art performance. Code and models are available at https://github.com/yuesongtian/DGL-GAN.

show abstract

SmoothGrad: removing noise by adding noise

Cited by 502 publications

References 16 publications

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

Topological Representations of Local Explanations

DGL-GAN: Discriminator Guided Learning for GAN Compression

Contact Info

Product

Resources

About