A5034127832- Cell Image Analysis Techniques
- Generative Adversarial Networks and Image Synthesis
- Genomics and Phylogenetic Studies
- Animal Behavior and Reproduction
- Marine animal studies overview
- Advanced Vision and Imaging
- Animal Vocal Communication and Behavior
- Modular Robots and Swarm Intelligence
- Ocular Surface and Contact Lens
- Coral and Marine Ecosystems Studies
- Planetary Science and Exploration
- Cephalopods and Marine Biology
- Domain Adaptation and Few-Shot Learning
- Insect and Arachnid Ecology and Behavior
- Primate Behavior and Ecology
- Astro and Planetary Science
- Plant and animal studies
Jet Propulsion Laboratory
2024
Max Planck Institute of Animal Behavior
2020-2021
University of Konstanz
2020-2021
Max Planck Society
2020
The University of Texas at Austin
2015-2017
Ice worlds are at the forefront of astrobiological interest because evidence subsurface oceans. Enceladus in particular is unique among icy moons there known vent systems that likely connected to a ocean, through which ocean water ejected space. An existing study has shown sending small robots into vents and directly sampling possible. To enable such mission, NASA’s Jet Propulsion Laboratory developing snake-like robot called Exobiology Extant Life Surveyor (EELS) can navigate Enceladus’...
Disappearing act Unlike coastal regions and reefs, the open ocean is mostly empty. Many fish species, nonetheless, spend most of their lives there. Such emptiness makes camouflage exceedingly difficult, so how does an organism hide in water filled with bouncing reflected light? Brady et al. show that some families have evolved skin reflects polarizes light, allowing them to blend into mirrorlike conditions more easily. These results help explain silvery coloration found sea-living across...
Predator evasion in the open ocean is difficult because there are no objects to hide behind. The silvery surface of fish plays an important role water camouflage. Various models have been proposed account for broadband reflectance by skin that involve one-dimensional variations arrangement guanine crystal reflectors, yet three-dimensional organization these platelets not well characterized. Here, we report and optical properties integumentary a marine fish, lookdown (Selene vomer). Our...
Significance To obtain and defend resources, animals often participate in contests that involve complex trade-offs between risk reward. Although it stands to reason invoke cognitive decision-making schemes analyze model these contests, such are hard verify behavioral experiments, they do not address a directly observable aspect of contests–the motion contestants space. We study the dynamics an orb-weaving spider, where males compete for mating opportunities bounded arena female’s web. show...
Apart from discriminative modeling, the application of deep convolutional neural networks to basic research utilizing natural imaging data faces unique hurdles. Here, we present decontextualized hierarchical representation learning (DHRL), designed specifically overcome these limitations. DHRL enables broader use small datasets, which are typical in most studies. It also captures spatial relationships between features, provides novel tools for investigating latent variables, and achieves...
S ummary Apart from discriminative models for classification and object detection tasks, the application of deep convolutional neural networks to basic research utilizing natural imaging data has been somewhat limited; particularly in cases where a set interpretable features downstream analysis is needed, key requirement many scientific investigations. We present an algorithm training paradigm designed specifically address this: decontextualized hierarchical representation learning (DHRL)....
Apart from discriminative models for classification and object detection tasks, the application of deep convolutional neural networks to basic research utilizing natural imaging data has been somewhat limited; particularly in cases where a set interpretable features downstream analysis is needed, key requirement many scientific investigations. We present an algorithm training paradigm designed specifically address this: decontextualized hierarchical representation learning (DHRL). By...