Neural Networks

Q1 Iris Dataset Classification via Neural Networks
Classify the IRIS data set via Artificial Neural Networks ANN
https://www.kaggle.com/azzion/iris-data-set-classification-using-neural-network/data
a.Compute the Confusion Matrix and Accuracy for the ANN
b.Compare the performance with a Naïve Bayes Classifier
Q2 Go through the Google Machine Learning Crash Course (free)
https://developers.google.com/machine-learning/crash-course
And write a paper describing the following exercises:
a.Explain the stochastic gradient descent algorithm for reducing loss
https://developers.google.com/machine-learning/crash-course/reducing-loss/gradient-descent
b.With the playground exercise explain the relationship between learning rate and convergence
https://developers.google.com/machine-learning/crash-course/reducing-loss/playground-exercise
c.Explain the anatomy of a neural network in your paper
https://developers.google.com/machine-learning/crash-course/introduction-to-neural-networks/anatomy
d.Review playground exercises on neural networks and explain in your paper
https://developers.google.com/machine-learning/crash-course/introduction-to-neural-networks/playground-exercises
Review the course in your paper and explain how it helps in understanding what neural networks are

Sample Solution

Assessment of obesity is done on the basis of BMI calculated by weight/height in m2 and plotting it on a BMI chart. Unfortunately this parameter cannot take into account the lean muscle mass of an individual. Muscular children may also have a higher BMI and racial/ethnic differences have been found in the fat content of individuals with the same BMI. On the other hand 25% children with a normal BMI have excess body fat. The risk of obesity related complications would be lower in children with higher muscle mass than in those with higher adiposity. Higher fat content and its distribution, especially central adiposity correlate better with the risk of obesity related complications. Hence Waist circumference may be a better parameter for predicting complications. Since measuring WC can be tedious for Pediatricians and most children with high BMI do have excess body fat, BMI should be used for assessing obesity.
BMI charts:
IAP Charts: BMI charts for Indian Children 5 to 18 years age were updated in 2015. The 23 and 27 adult equivalent cut offs lines (for risk of overweight and obesity, respectively) are similar to the IOTF cut-offs and are more appropriate for use in Asian children since they are known to have more adiposity and increased cardio-metabolic risk at a lower BMI (2). Hence it is preferable to use Indian IAP charts for our population 5-18years, WHO BMI charts from 2-5yrs age and weight for height charts by WHO for children <2 yrs age.
CDC charts: Children and adolescents ≥2 years of age are diagnosed as overweight if the BMI is ≥85th percentile but <95th percentile and obese if the BMI is ≥95th percentile for age

This question has been answered.

Get Answer