Hadoop data mining swiss army knife
Alexandre Dutra Cançado, Bertrand Dechoux, Pablo Lopez
Map/reduce: Hadoop
- record (key/value) based
- map/group-by/reduce
- distributed batch
Hive
- load data in table
- sql-like syntax to split lines / do processing (sql can become complex)
- metastore / HCatalog
- external tooling (Tableau)
- partitionng not trivial
- new language for map/reduce
- procedural
Cascading:
- java api
- procedural
Enterprise Concurrency - practical solutions with JSR236
Alex Heusingfeld refs
- http://www.slideshare.net/FredRowe/jsr-236-concurrency-utils-for-ee-presentation-for-java-onesf2013-con7948jsr236
- http://jcp.org/en/jsr/detail?id=236
- https://blogs.oracle.com/arungupta/entry/concurrency_utilities_for_java_ee
Taming Drones: How Java Controls the Uprising of the Drone Force
Eva Veenstra-Kazakov, Timon Veenstra
"specialists at low cost" e.g.
- rainfall damage - aerial overview
- biomass estimation -> AgroSense tool
- track spraying
selled as toy: quadcopter Parrot AR.Drone 2.0
- library for lowlevel access
- javafx app (wifi): instructions / data
- inside / outside options
- sensors: cameras, sonar, speed ...
- API AgroSense builds on top of ARDRoneForP5
autopilot through java: very experimental
- extra raspberry pi (reuse existing usb connector)
- gps + big antenna - accurate to 2m -> correct with gsm signal (RTK-GPS)
- challenges: wind, 30 min fly time, mandatory wifi connection, battery, charging, follow tractor...
refs:
No comments:
Post a Comment