Oral Academic Presentation about Our Work ACSleuth
Date:
We presented our team’s work, ACSleuth, discussing in detail its implementation specifics, applicable tasks, and the limitations of the algorithm.
In this talk, we introduced the unique features of ACSleuth in the task of anomalous cell detection. Firstly, unlike other uncertainty-based methods, ACSleuth exhibits higher tolerance to biases between training and testing datasets. Secondly, ACSleuth is not limited to detecting cancer cells but can identify any samples different from the training dataset. Finally, we discussed the future direction of our team’s work, which involves extending the framework of ACSleuth to enable spatial transcriptomics data anomaly region detection via multimodal learning.