The elephant in the room.

By J. Dunning. 2013

537 citations recorded by the source corpus.

Subfields: Cardiology and Cardiovascular Medicine, Emergency Medical Services, Emergency Medicine, Philosophy

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cardiology and Cardiovascular Medicine

In Emergency Medical Services

In Emergency Medicine

In Philosophy

Citations in the corpus

Shortcut learning in deep neural networks (2020, background): “DNNs achieve super-human performance recognising objects, but even small invisible changes [8] or a different background context [9, 10] can completely derail predictions.”
Strike (With) a Pose: Neural Networks Are Easily Fooled by Strange Poses of Familiar Objects (2018, background): “2 (2)Note that the sensitivity of classifiers and object detectors to 2D translations has been observed in concurrent work [35, 12, 56, 6].”
An Analysis of the Adaptation Speed of Causal Models (2020, background): “Modern machine learning methods can fail surprisingly when the test distribution differ from the training distribution (Rosenfeld et al., 2018).”
Why do deep convolutional networks generalize so poorly to small image transformations? (2018, background): “, 2018) and detection (Rosenfeld et al., 2018) along with studies suggesting that with small CNNs and the MNIST data, data augmentation is the main feature affecting CNN invariance (Kauderer-Abrams, 2017).”
Motivating the Rules of the Game for Adversarial Example Research (2018, background): “For example, Rosenfeld [100] showed that randomly moving around an object in the background may occasionally cause other objects to go undetected or become misclassiﬁed.”
Adversarial Examples Are a Natural Consequence of Test Error in Noise (2019, background): “…they lack robustness to different types of blur, pixelation, and changes in brightness (Hendrycks & Dietterich, 2018), lack robustness to random translations of the input (Azulay & Weiss, 2018), and even make errors when foreign objects are inserted into the ﬁeld of view (Rosenfeld et al., 2018).”
Improving Robustness Without Sacrificing Accuracy with Patch Gaussian Augmentation (2019, background): “For example, neural networks are sensitive to small translations and changes in scale [2], blurring and additive noise [10], small objects placed in images [41], and even different images from a similar distribution of the training set [40, 39].”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “First, several works have shown that object classifiers often rely on the background rather than the object to make its classification (Rosenfeld et al., 2018; Shetty et al., 2019; Xiao et al., 2020).”
Unsupervised Visuomotor Control through Distributional Planning Networks (2019, background): “First, object poses are not readily accessible and pre-trained object detectors struggle without fine-tuning with data collected in the robot’s domain [39].”
Deep Nets: What have They Ever Done for Vision? (2018, background): “an elephant in a room (Rosenfeld et al., 2018).”
The Role of Leadership in a Digitalized World: A Review (2019, methodology): “To answer our research questions, we use a mixed method approach, that involves both quantitative research through standard databases and qualitative coding (Crossan and Apaydin, 2010; Peteraf et al., 2013; Zupic and Čater, 2015).”
REVISE: A Tool for Measuring and Mitigating Bias in Visual Datasets (2020, background): “…2019; Ouyang et al., 2016; Yang et al., 2014; Salakhutdinov et al., 2011; Buda et al., 2017; Liu et al., 2009), object contexts (Choi et al., 2012; Rosenfeld et al., 2018), object sub-categories (Zhu et al., 2014), scenes (Zhou et al., 2017), gender (Kay et al., 2015), gender contexts (Zhao et…”
Affinity and Diversity: Quantifying Mechanisms of Data Augmentation (2020, methodology): “It has been used to measure robustness to corruptions of images that do not change their semantic content (Azulay & Weiss, 2018; Dodge & Karam, 2017; Ford et al., 2019; Hendrycks & Dietterich, 2018; Rosenfeld et al., 2018; Yin et al., 2019).”
Delving Deep into the Generalization of Vision Transformers under Distribution Shifts (2021, background): “However, previous works have demonstrated that backgrounds may dominate in prediction [2, 26], which is undesirable to us.”
Gradient Starvation: A Learning Proclivity in Neural Networks (2020, background): “In many cases, state-of-the-art neural networks appear to focus on low-level superficial correlations, rather than more abstract and robustly informative features of interest [16, 88, 40, 68, 30].”
Tradeoffs in Data Augmentation: An Empirical Study (2021, methodology): “This same metric has been used as a measure of a model’s robustness to image corruptions that do not change images’ semantic content (Azulay & Weiss, 2018; Dodge & Karam, 2017; Ford et al., 2019; Hendrycks & Dietterich, 2019; Rosenfeld et al., 2018; Yin et al., 2019).”
Noise or Signal: The Role of Image Backgrounds in Object Recognition (2020, background): “Indeed, prior work has shown that models may use backgrounds in classiﬁcation [Zha+07; RSG16; ZXY17; RZT18; Bar+19; SSF19; Sag+20], and suggests that even human vision makes use of image context for scene and object recognition [Tor03].”
Counterfactual Generative Networks (2021, background): “, 2013), or failing to recognize known objects in unfamiliar contexts (Rosenfeld et al., 2018) or from unseen poses (Alcorn et al.”
EDF: Ensemble, Distill, and Fuse for Easy Video Labeling (2018, background): “Previous work shows that slight modiﬁcations to a frame can have signiﬁcant non-local impacts on object detection [23].”
Bag of Freebies for Training Object Detection Neural Networks (2019, methodology): “[18], we focus on the natural co-occurrence object presentations which play signiﬁcant roles in object detection.”