Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining [PDF]
Mainstream 3D representation learning approaches are built upon contrastive or generative modeling pretext tasks, where great improvements in performance on various downstream tasks have been achieved.
Zekun Qi +6 more
semanticscholar +1 more source
An Aligned Rank Transform Procedure for Multifactor Contrast Tests [PDF]
Data from multifactor HCI experiments often violates the assumptions of parametric tests (i.e., nonconforming data). The Aligned Rank Transform (ART) has become a popular nonparametric analysis in HCI that can find main and interaction effects in ...
Lisa Elkin +3 more
semanticscholar +1 more source
Exploring Cross-Image Pixel Contrast for Semantic Segmentation [PDF]
Current semantic segmentation methods focus only on mining "local" context, i.e., dependencies between pixels within individual images, by context-aggregation modules (e.g., dilated convolution, neural attention) or structure-aware optimization criteria (
Wenguan Wang +5 more
semanticscholar +1 more source
Dimethyl 3,3′-[(4-nitrophenyl)methylene]bis(1H-indole-2-carboxylate) ethanol hemisolvate
There are two main molecules in asymmetric unit of the title compound, C27H21N3O6·0.5C2H5OH. In both, the indole ring systems are approximately perpendicular to each other, at dihedral angles of 69.3 (5) and 82.8(4)°. In the crystal, molecules are linked
Yu-Long Li +6 more
doaj +1 more source
Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlations [PDF]
Spurious correlations pose a major challenge for robust machine learning. Models trained with empirical risk minimization (ERM) may learn to rely on correlations between class labels and spurious attributes, leading to poor performance on data groups ...
Michael Zhang +4 more
semanticscholar +1 more source
Momentum Contrast for Unsupervised Visual Representation Learning [PDF]
We present Momentum Contrast (MoCo) for unsupervised visual representation learning. From a perspective on contrastive learning as dictionary look-up, we build a dynamic dictionary with a queue and a moving-averaged encoder. This enables building a large
Kaiming He +4 more
semanticscholar +1 more source
Attentional Local Contrast Networks for Infrared Small Target Detection [PDF]
To mitigate the issue of minimal intrinsic features for pure data-driven methods, in this article, we propose a novel model-driven deep network for infrared small target detection, which combines discriminative networks and conventional model-driven ...
Yimian Dai +3 more
semanticscholar +1 more source
7,8-Dimethyl-11H-indeno[1,2-b]quinoxalin-11-one
In the title compound, C17H12N2O, the mean planes of the indene ring and quinoxaline system (r.m.s. deviations = 0.0131 and 0.0082 Å) are approximately parallel to one another, making a dihedral angle of 1.2 (5)°.
Lin Chen, Jin Hu, Hong-Shun Sun
doaj +1 more source
Evidence-based nutritional guidelines: What is meant by trustworthy recommendations
The article aims to describe the characteristics of trustworthy recommendations as well as standards for trustworthy guidelines published by the Institute of Medicine and tools that can be used for quality assessment.
Joanna Zając +2 more
doaj +1 more source
Diethyl 3,3′-[(3-fluorophenyl)methylene]bis(1H-indole-2-carboxylate)
In the title compound, C29H25FN2O4, the mean planes of the indole ring systems are approximately perpendicular to one another [dihedral angle = 88.3 (4)°]. The benzene ring is twisted with respect to the indole ring systems by 49.8 (5) and 77.6 (3)°.
Hong Jiang +7 more
doaj +1 more source

