What does it mean for animals to exhibit complex social behaviors in their daily lives?How do birds communicate information silently when they cluster and take off synchronously?Why do fish change their swimming patterns when threatened?Exploring the neural mechanisms and behavioral patterns behind these questions can help understand the deeper implications of animal social language.
The development of artificial intelligence technology has not only greatly helped scientists analyze and understand biological behavior, but also provided a revolutionary perspective for neuroscience research.
The latest research results of the research team of Shenzhen Institute of Advanced Technology of Chinese Academy of Sciences, Shenzhen Institute of Cognition and Brain Diseases, Shenzhen University of Technology (Shenzhen) were listed in the top international academic journal Nature-machine intelligence.The research team applied artificial intelligence technology to neuroscience research, and proposed a small-sample learning computing framework model, Social Behavior Atlas (SBeA), which realized unlabeled, high-precision 3D pose estimation, zero-sample identification and refined social behavior classification.
Artificial Intelligence Helps Neurological Diseases From Laboratory, Natural Environment to Clinical Medicine Series Research Schematic Diagram
"Multi-animal behavior quantification is key to interpreting animal social behavior and has broad applications in neuroscience and ecology.Trenton Jerde, senior editor of the journal Nature-Machine Intelligence, gave the study a very high rating.
AI Empowerment, Zero Labeling Accurate Identification of Animals
In the free social state, animals with different disease models often have different disease states and need to locate related behavioral phenotypes one by one.However, the animal's appearance is completely covered with fur, making it difficult for even experienced professional experimenters to distinguish the identity and state of each animal in a social scene.
In recent years, the widespread use of AI technology in traditional behavioral research has led to the development of computational neurobehavioral science, and AI animal behavior tracking technologies such as DeepLabCut, SLEAP, and MoSeq are becoming important research tools.
However, when analyzing multiple animal targets and animal free social behavior, the above technology still fails to achieve problems such as large-scale data labeling and continuous tracking of identity accuracy.At present, there are no truly effective tools available for detecting animal behavior data, and most animal social behavior research is still in the traditional three-box behavioral experiment stage.
At present, the accuracy of artificial intelligence recognition images has greatly exceeded that of the human eye.In this study, the research team applied AI technology to animal identification and proposed a two-way transfer learning computing framework to achieve zero-sample multi-animal social identification without labeling animal identification data in advance.
In non-social situations, it is very simple to distinguish the identities of each animal. These models can transfer animal identification information to multi-animal social situations, enabling knowledge sharing and 'brain complement' between different models.The paper's corresponding author, Wei Pengfei, said.
Universal application across species to aid neurological disease research
In the previous research, Weipengfei's team developed a framework for fine behavior analysis of single animals and miniaturized free-moving eye tracking equipment.
However, these methods can only be applied in single-animal experimental scenarios, limiting research expansion from laboratory to clinical care."The flying presentation.Since the end of 2020, the research team has conceived and tried to extend unsupervised, hierarchical dynamic behavioral decomposition strategies to social behavioral scenarios.
After three years of unremitting efforts, the cross-disciplinary team led by Wei Pengfei solved a large number of artificial intelligence technology problems and completed the detailed behavior analysis of multi-animal targets.
SBeA technology developed by the research team has realized parallel, dynamic and hierarchical decomposition of animal social behavior, adaptive and unsupervised clustering of animal social behavior characteristics such as mice, rats, birds, dogs, and nonhuman primates.
This method does not need to define social behavior categories in advance, can help identify new and undefined social behavior differences, and can identify free social behavior phenotypes that are difficult to obtain in classic social experimental paradigms.
"Over the course of the study, we found an interesting phenomenon in which mice, like humans, proactively 'care' about other mice, especially model mice with mental illness."The system found that normal mice may have a higher preference for autism model mice.The first author of the paper, Hanning, a doctoral student at Shenzhen Advanced Institute, introduced, which provides a new idea for the team to further analyze the deeper meaning of animal social behavior.
In addition to mice, the method is also suitable for calculating accurate 3D social postures, identities, and fine social modules for birds and domestic dogs, with the potential for cross-species applications.Among them, the research work of domestic dogs is mainly carried out in cooperation with academician Zhang Yaping of Kunming Institute of Zoology of Chinese Academy of Sciences, Wang Guodong and Li Jing of Kunming Police Dog Base.
Luo Minmin, a professor at Tsinghua University and co-director of the Center for Brain Science and Brain-like Studies in Beijing, said that Social Behavior Atlas can be seen as a "magnifier" in behavioral research.
"In recent years, computational neurobehavioral developments have revolutionized classical behavioral paradigms, enabling behavioral observations from laboratory environments to natural environments.In the future, AI-enabled neuroscience research will further deepen the understanding of big data physiological parameters from animal models to clinical medicine, providing guidance for more accurate and personalized non-invasive neural regulation."Flying said.