@Slash0BZ found out during his investigations for #101 that the current implementation of Support Vector Machine causes an increase in Heap memory usage linearly with the number of iterations of training. Need to investigate this.
Additionally, we tried commenting out the LibLinear train method calls but the issue still seems to happen.
@Slash0BZ found out during his investigations for #101 that the current implementation of Support Vector Machine causes an increase in Heap memory usage linearly with the number of iterations of training. Need to investigate this.
Additionally, we tried commenting out the LibLinear train method calls but the issue still seems to happen.
Memory usage log: http://gceasy.io/my-gc-report.jsp?p=c2hhcmVkLzIwMTcvMDIvMjcvLS1tZW1vcnkubG9nLS0xOC0yMS0zOQ==