Here is a summary of the ML challenges to reach mass market with APIs:
- Simplify the preprocessing (data cleaning, features extraction & selection-> 90% work) & integration into a mining or ML (10%) & http:
//scikit-learn.org (opensource project supported by google & INRIA research group) - Simplification of data visualization
- Simplification of semi supervised tagging (reduce the tagging/labelling effort)
- Simplification of parameter selection (model included): Hyperopt: A Python library for optimizing machine learning algorithms; SciPy 2013 - YouTube
- For services in the cloud, the biggest show stopper is data transfer (way to slow) & confidentiality.
Players to watch:
PredictionIO : open source machine learning server
Apache Mahout: Scalable Machine Learning and data-mining
No comments:
Post a Comment