Learning Speaker Representation with Semi-supervised Learning approach for Speaker Profiling