In the paper titled Supervised Learning Algorithm: SVM with Advanced Kernel to classify Lower Back Pain, the authors Mittal Bhatt, Vishal Dahiya & Arvind Singh have introduced an approach to classify the condition of Spine and near disorders which can lead to Chronic Lower Back Pain using a kernel designed in Support Vector Machine. It is also shown that use of ANNs (Artificial Neural Network) are more reliable and accurate in the field of medical science as it simulates human behaviour and also possesses abilities of generalization and learning. Paper shows an Expert System which is made as an application of AI with a knowledge domain fed into it. This Expert System being interactive has roles like diagnosing, interpreting, predicting, and instructing is effective in medical sciences for classification of LBP. Author has emphasised on the fact that SVM finds the hyperplane which separates input elements with maximal margin in the n dimensional space by mapping training vectors product known as kernel function. For the efficient performance of SVM the choice of kernel function is very important and thus Author has discussed different kernel functions that can be used in SVM so as to make an effective Expert System. Different kernel functions include Polynomial Kernel Function, Gaussian Radial Basis Function, Exponential Radial Basis Function, Multi-Layer perceptron and made an SVM algorithm for them and applied on different datasets to compare accuracy. The newly designed weighted kernel function is then implemented on the same dataset which helps in knowing which attributes are important in classification of LBP more accurately which in turn will make the Expert system better and help medical practitioners to classify Lower back Pain easily.

Important Sentences - 

• Artificial Neural Network (ANN) exhibits the behavior with integrity of correctness which makes it most reliable in medical science.
• An Expert System (ES) is one of the application areas of Artificial Intelligence, which have expert knowledge of a particular domain and it uses this knowledge to respond properly.
• An ES can take mainly four interactive roles, diagnosing, interpreting, predicting and instructing, n  number of various application areas like medical field, education, agriculture etc.
• ANN simulates human behavior, it possesses great capabilities of learning and generalization.
• SVM finds the hyperplane that separates input elements with maximal margin in the              

n dimensional space by mapping training vectors product known as kernel function.

• For the efficient performance of SVM the choice of kernel function is very important.
• The kernel function is a dot product of variables in the input space and maps it to the output space. That is, the kernel defines the inner product in feature space.
• The different types of kernel functions are listed below:

1. The simplest kernel function defined in SVM as inner product of vectors:

2. Polynomial Kernel Function: It is a renowned method for non-linear modeling. The second kernel is usually preferable as it avoids problems with the hessian becoming Zero.

3. Gaussian Radial Basis Function: It is most commonly with a Gaussian form.

4. Exponential Radial Basis Function: its output is a piecewise linear solution which can be useful when discontinuities are acceptable.

5. Multi-Layer Perceptron: With a single hidden layer, a valid kernel representation is as follows:

• The newly designed weighted kernel function is implemented for different datasets and it is observed that the attributes, sacral_slope, pelvic_radius, degree_spondylolisthesis, pelvic_slope affects most on classification of lower back pain more accurately.

• Thus the designed weighted kernel is used to identify the attributes that most affects the classification of spine condition.