Kernel Based Extreme Learning Machine in Identifying Dermatological Disorders

Extreme Learning Machine (ELM)has gained importance among various learning algorithms particularly related to multi classification problems. In this paper, kernel-based ELM is used to identify common skin disorders such as Bacterial Infection, Fungal Infection, Eczema and Scabies. A proper diagnosis of these diseases at primary stage is very essential to prevent future complications. The various kernel functions like Radial Basis Function, Polynomial kernel, Exponential Chi-Square kernel with ELM are applied on the skin dataset. In our study we measure accuracy and leaning time obtained using these kernels with ELM and Support Vector Machine (SVM). We also analyze our dataset on Conventional Single Layer feed Forward Network (SLFN). A comparative study of accuracy and learning time of all these learning algorithms is made and it is observed that ELM gives optimum learning time with good classification accuracy using exponential chi-square kernel.