|
- New statistical techniques for the processing of astronomical
data: time series, images, low number statistics for high energy
photons, heteroskedastic data, non-detections...
-
Challenges in the data mining of astronomical
databases:
- the class imbalance in training sets or how to define priors
- robust preprocessing for supervised/unsupervised
classification
- robust inference with heterogeneous datasets, or how to combine
(wildly different) observations, models, priors, etc in a
training/test set
- error
propagation
- The challenge of petabyte size databases: scalability, parallel
computing, accuracy.Geometric data organization, sky indexing for
efficient data retrieval, intelligent access to petabyte size
databases
- Knowledge Discovery in
astronomical archives: outlier detection, new object types, parametric
inference, model fitting and model selection, etc.
- Combining the
classical domain knowledge approach with machine learning
techniques.
- Global approaches for
global datasets. The Galaxy zoo and the Universe zoo.
- The Virtual Observatories, Data Mining and Astrostatistics: software, standards, protocols...
|