distributed system for the K-nearest-neighbors algorithm, it essentially implements map reduce with a depth level of one.
Structure:
- Master starts up and waits for a number of consumers to connect.
- Client sends to master feature vectors
- Master broadcasts to consumers
- Consumers compute euclidean distance, send top k results to aggregator
- Aggregator compiles the results and returns the top k
- Aggregator finds most frequent and sends to master ID and category
- Master sends results back to consumer
- Consumer writes results to file
To compile: make
To run:
- Master: make runMaster
- Consumer: make runConsumer
- Client: make runClient
- Accumulator: make runAcc
To clean:
- make clean
To edit the arguments
- open pom.xml
- Go down to the profiles section where you'll see profiles for the master, consumer, client, and accumulator. In each profile, there's a list of tags with the argument description next to it. Add your argument here.