Share This Author
The psychometric function: I. Fitting, sampling, and goodness of fit
An integrated approach to fitting psychometric functions, assessing the goodness of fit, and providing confidence intervals for the function’s parameters and other estimates derived from them, for the purposes of hypothesis testing is described.
ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
- Robert Geirhos, Patricia Rubisch, Claudio Michaelis, M. Bethge, Felix Wichmann, Wieland Brendel
- 27 September 2018
It is shown that ImageNet-trained CNNs are strongly biased towards recognising textures rather than shapes, which is in stark contrast to human behavioural evidence and reveals fundamentally different classification strategies.
The psychometric function: II. Bootstrap-based confidence intervals and sampling
The present paper’s principal topic is the estimation of the variability of fitted parameters and derived quantities, such as thresholds and slopes, and introduces improved confidence intervals that improve on the parametric and percentile-based bootstrap confidence intervals previously used.
The psychometric function: I
Heterocyclic nitrogen-containing peroxides are prepared by reacting a tertiary alkyl hydroperoxide with a nitrogen heterocycle substituted by cycloalkene such as 1-(4-morpholinyl)cyclohexene. The…
Shortcut Learning in Deep Neural Networks
A set of recommendations for model interpretation and benchmarking is developed, highlighting recent advances in machine learning to improve robustness and transferability from the lab to real-world applications.
Generalisation in humans and deep neural networks
- Robert Geirhos, Carlos R. Medina Temme, Jonas Rauber, H. Schütt, M. Bethge, Felix Wichmann
- Computer ScienceNeurIPS
- 27 August 2018
The robustness of humans and current convolutional deep neural networks on object recognition under twelve different types of image degradations is compared and it is shown that DNNs trained directly on distorted images consistently surpass human performance on the exact distortion types they were trained on.
Inference for psychometric functions in the presence of nonstationary behavior.
Monte Carlo simulations are used to show that violations of these assumptions can result in underestimation of confidence intervals for parameters of the psychometric function and a simple adjustment of the confidence intervals is presented that corrects for the underestimation almost independently of the number of trials and the particular type of violation.
Painfree and accurate Bayesian estimation of psychometric functions for (potentially) overdispersed data
Bayesian inference for psychometric functions.
This study proposes the use of Bayesian inference to extract the information contained in experimental data to estimate the parameters of psychometric functions and describes how a Markov chain Monte Carlo method can be used to generate samples from the posterior distribution over parameters.
Transcranial magnetic stimulation in the visual system. I. The psychophysics of visual suppression
- T. Kammer, K. Puls, H. Strasburger, N. Hill, Felix Wichmann
- PsychologyExperimental Brain Research
The finding that at higher TMS intensities, threshold elevation occurs even with shorter SOAs, this suggests lasting inhibitory processes as masking mechanisms, contradicting the assumption that the phosphene as excitatory equivalent causes masking.