In the present scenario, one of the dangerous disease is cancer. It spreads
through blood or lymph to other location of the body, it is a set of cells
display uncontrolled growth, attack and destroy nearby tissues, and
occasionally metastasis. In cancer diagnosis and molecular biology, a
utilized effective tool is DNA microarrays. The dominance of this technique
is recognized, so several open doubt arise regarding proper examination of
microarray data. In the field of medical sciences, multicategory cancer
classification plays very important role. The need for cancer classification
has become essential because the number of cancer sufferers is increasing. In
this research work, to overcome problems of multicategory cancer
classification an improved Extreme Learning Machine (ELM) classifier is used.
It rectify problems faced by iterative learning methods such as local minima,
improper learning rate and over fitting and the training completes with high
speed.