01 January 1970 0 6K Report

I want to process remote sensing images by implementing U-Net neural networks on Apache Spark (Pyspark). Both training and inference processes are required to be run on Spark. Well, I have no good work experience on Spark but after googling I found several frameworks/libraries to choose from: 1. Spark’s MLlib 2. TensorflowOnSpark 3. Deep Learning Pipelines (don't support distributed learning) 4- Horovod 5- BigDL

6- Apache SystemML I got confused about what to choose from in my case! Which one is suitable? Isn't there any better alternative?

More Orwa Kassab's questions See All
Similar questions and discussions