Felix A. Wichmann
YOU?
Author Swipe
View article: PathMR: Multimodal Visual Reasoning for Interpretable Pathology Diagnosis
PathMR: Multimodal Visual Reasoning for Interpretable Pathology Diagnosis Open
Deep learning based automated pathological diagnosis has markedly improved diagnostic efficiency and reduced variability between observers, yet its clinical adoption remains limited by opaque model decisions and a lack of traceable rationa…
View article: Quantifying Uncertainty in Error Consistency: Towards Reliable Behavioral Comparison of Classifiers
Quantifying Uncertainty in Error Consistency: Towards Reliable Behavioral Comparison of Classifiers Open
Benchmarking models is a key factor for the rapid progress in machine learning (ML) research. Thus, further progress depends on improving benchmarking metrics. A standard metric to measure the behavioral alignment between ML models and hum…
View article: Estimating the contribution of early and late noise in vision from psychophysical data
Estimating the contribution of early and late noise in vision from psychophysical data Open
Human performance in psychophysical detection and discrimination tasks is limited by inner noise. It is unclear to what extent this inner noise arises from early noise (e.g., in the photoreceptors) or from late noise (at or immediately pri…
View article: How Aligned are Different Alignment Metrics?
How Aligned are Different Alignment Metrics? Open
In recent years, various methods and benchmarks have been proposed to empirically evaluate the alignment of artificial neural networks to human neural and behavioral data. But how aligned are different alignment metrics? To answer this que…
View article: Standard models of spatial vision mispredict edge sensitivity at low spatial frequencies
Standard models of spatial vision mispredict edge sensitivity at low spatial frequencies Open
One well-established characteristic of early visual processing is the contrast sensitivity function (CSF) which describes how sensitivity varies with the spatial frequency (SF) content of the visual input. The CSF prompted the development …
View article: Comparing supervised learning dynamics: Deep neural networks match human data efficiency but show a generalisation lag
Comparing supervised learning dynamics: Deep neural networks match human data efficiency but show a generalisation lag Open
Recent research has seen many behavioral comparisons between humans and deep neural networks (DNNs) in the domain of image classification. Often, comparison studies focus on the end-result of the learning process by measuring and comparing…
View article: Neither hype nor gloom do DNNs justice
Neither hype nor gloom do DNNs justice Open
Neither the hype exemplified in some exaggerated claims about deep neural networks (DNNs), nor the gloom expressed by Bowers et al. do DNNs as models in vision science justice: DNNs rapidly evolve, and today's limitations are often tomorro…
View article: Seeing the future of progressive glasses: a new perceptual approach to spectacle lens design
Seeing the future of progressive glasses: a new perceptual approach to spectacle lens design Open
The eye’s natural aging influences our ability to focus on close objects. Without optical correction, all adults will suffer from blurry close vision starting in their 40s. In effect, different optical corrections are necessary for near an…
View article: Feature Visualizations do not sufficiently explain hidden units of Artificial Neural Networks
Feature Visualizations do not sufficiently explain hidden units of Artificial Neural Networks Open
Artificial Neural Networks (ANNs) have been proposed as computational models of the primate ventral stream, because their performance on tasks such as image classification rivals or exceeds human baselines. But useful models should not onl…
View article: Is edge sensitivity more than contrast sensitivity?
Is edge sensitivity more than contrast sensitivity? Open
How does the human visual system extract relevant information from the pattern of light on the retinae? The psychophysical study of early vision has led to important insights in the initial processing characteristics of human vision, and w…
View article: Measuring lightness constancy with varying realism
Measuring lightness constancy with varying realism Open
In everyday life, surface colours typically appear constant under different lighting despite the (potentially) dramatic changes in luminance—human observers are (nearly) colour constant. Failures of colour constancy using simple artificial…
View article: Psychophysical scale of optical distortions of multifocal spectacle lenses
Psychophysical scale of optical distortions of multifocal spectacle lenses Open
Multifocal lenses have regions of different optical power, correcting far and near vision for presbyopes with one pair of glasses. Optical distortions are an unavoidable side effect of those lenses. From previous research and reports of sp…
View article: The developmental trajectory of object recognition robustness: Children are like small adults but unlike big deep neural networks
The developmental trajectory of object recognition robustness: Children are like small adults but unlike big deep neural networks Open
In laboratory object recognition tasks based on undistorted photographs, both adult humans and deep neural networks (DNNs) perform close to ceiling. Unlike adults', whose object recognition performance is robust against a wide range of ima…
View article: Are Deep Neural Networks Adequate Behavioural Models of Human Visual Perception?
Are Deep Neural Networks Adequate Behavioural Models of Human Visual Perception? Open
Deep neural networks (DNNs) are machine learning algorithms that have revolutionised computer vision due to their remarkable successes in tasks like object classification and segmentation. The success of DNNs as computer vision algorithms …
View article: Are Deep Neural Networks Adequate Behavioral Models of Human Visual Perception?
Are Deep Neural Networks Adequate Behavioral Models of Human Visual Perception? Open
Deep neural networks (DNNs) are machine learning algorithms that have revolutionized computer vision due to their remarkable successes in tasks like object classification and segmentation. The success of DNNs as computer vision algorithms …
View article: Estimating the perceived dimensionality of psychophysical stimuli using a triplet accuracy and hypothesis testing procedure
Estimating the perceived dimensionality of psychophysical stimuli using a triplet accuracy and hypothesis testing procedure Open
In vision research we are often interested in understanding the mapping of complex physical stimuli to their perceptual dimensions. This mapping can be explored experimentally with multi-dimensional psychophysical scaling. One fruitful app…
View article: The bittersweet lesson: data-rich models narrow the behavioural gap to human vision
The bittersweet lesson: data-rich models narrow the behavioural gap to human vision Open
A major obstacle to understanding human visual object recognition is our lack of behaviourally faithful models. Even the best models based on deep learning classifiers strikingly deviate from human perception in many ways. To study this de…
View article: Using an Odd-One-Out Design Affects Consistency, Agreement and Decision Criteria in Similarity Judgement Tasks Involving Natural Images.
Using an Odd-One-Out Design Affects Consistency, Agreement and Decision Criteria in Similarity Judgement Tasks Involving Natural Images. Open
Recently, similarity judgement tasks have been employed to estimate the perceived similarity of natural images (Hebart, Zheng, Pereira, & Baker, 2020). Such tasks typically take the form of triplet questions in which participants are prese…
View article: Estimating the perceived dimension of psychophysical stimuli using triplet accuracy and hypothesis testing
Estimating the perceived dimension of psychophysical stimuli using triplet accuracy and hypothesis testing Open
Vision researchers are interested in mapping complex physical stimuli to perceptual dimensions. Such a mapping can be constructed using multidimensional psychophysical scaling or ordinal embedding methods. Both methods infer coordinates th…
View article: The developmental trajectory of object recognition robustness: children are like small adults but unlike big deep neural networks
The developmental trajectory of object recognition robustness: children are like small adults but unlike big deep neural networks Open
In laboratory object recognition tasks based on undistorted photographs, both adult humans and Deep Neural Networks (DNNs) perform close to ceiling. Unlike adults', whose object recognition performance is robust against a wide range of ima…
View article: Deep neural models for color classification and color constancy
Deep neural models for color classification and color constancy Open
Color constancy is our ability to perceive constant colors across varying illuminations. Here, we trained deep neural networks to be color constant and evaluated their performance with varying cues. Inputs to the networks consisted of two-…
View article: Trivial or impossible -- dichotomous data difficulty masks model differences (on ImageNet and beyond)
Trivial or impossible -- dichotomous data difficulty masks model differences (on ImageNet and beyond) Open
"The power of a generalization system follows directly from its biases" (Mitchell 1980). Today, CNNs are incredibly powerful generalisation systems -- but to what degree have we understood how their inductive bias influences model decision…
View article: Trivial or impossible -- dichotomous data difficulty masks model\n differences (on ImageNet and beyond)
Trivial or impossible -- dichotomous data difficulty masks model\n differences (on ImageNet and beyond) Open
"The power of a generalization system follows directly from its biases"\n(Mitchell 1980). Today, CNNs are incredibly powerful generalisation systems --\nbut to what degree have we understood how their inductive bias influences model\ndecis…
View article: The developmental trajectory of object recognition robustness: comparing children, adults, and CNNs
The developmental trajectory of object recognition robustness: comparing children, adults, and CNNs Open
Core object recognition refers to the ability to rapidly recognize objects in natural scenes across identity-preserving transformations, such as variation in perspective, size or lighting. In laboratory object recognition tasks using 2D im…