You would like to develop software that can detect 5 different animals from their pictures. These 5 animals include dog,

[answerhappygod]

You Would Like To Develop Software That Can Detect 5 Different Animals From Their Pictures These 5 Animals Include Dog 1 (40.88 KiB) Viewed 44 times

You would like to develop software that can detect 5 different animals from their pictures. These 5 animals include dog, cat, horse, pig, and cow. You decide to create the software using machine learning. 01. You want to formulate your problem as Tom Mitchell's Learning Problem. Fill the blank. Component Your Learning Problem Task Performance Measure Experience Q2. Instead of using pictures directly at first, you decide to measure the following characteristics of each animal. You managed to collect 50 samples per animal type. Characteristics Description Body Length (inch) From head to tail Body Height (inch) From head to feet Body (Fur) Color Red, Orange, Yellow, Green, Blue, Purple, Brown, Black, White, Pink Front Toes Number of Toes on a Front Leg Back Toes Number of Toes on a Back Leg These are some samples from your data. Back Toes 4 4 Sample Body Length Body Height 1 20 11 2 22 12 3 18 9 4 96 66 5 65 36 6 98 65 Please answer the following questions. Body Color Brown Yellow Black White Pink Red Front Toes 5 5 5 1 4 4 Animal Type Dog Dog Cat Horse Pig Cow 1 4 2 2 Q2-1. Convert non-numerical attributes to numerical ones. Q2-2. Write vector forms of samples from 1 to 5 excluding Animal Type. You need to declare which dimension represents which feature. Q3. You would like to create a machine learning model using the dataset from 02. Your goal is to achieve at least 85% of accuracy with the model. Please answer the following questions. 03-1. How many inputs does this model have? List all feature names. Q3-2. What are the label values for this model? List all label values. 03-3. The below table is a template for your model description. Each row represents a layer in the model. You do not need to fill a row if your model does not have the corresponding layer. Design the smallest model for this task. Number of Neurons Type of Layer (Dense or Softmax) Activation Function (Sigmoid, Bebe N/A if no need) Layer (Start from 1 next to Input) 1 2 3

03-4. How many model parameters (including all weights and bias values) does this model have? 03-5. What type of loss function do you need to use for this model? Q3-6. You trained your model above and got the following loss and accuracy curve after 30 epochs. Training loss Training accuracy 0.25 Validation loss Validation accuracy 65% Loss Accuracy 0.01 1 30 1 Epoch 30 Epoch Describe the current status of the model and what do you need to do to improve the model to meet your need. 03-7. Based on the discussion in Q3-6, update your model below. Layer Type of Layer Number of Neurons Activation Function (Start from 1 (Dense or Saitmax) (Sigmoid, ReLU, N/A if no need) next to Input) 1 2 3 03-8. You trained your model above and got the following loss and accuracy curve after 30 epochs. Training loss Training accuracy 0.25 Validation loss Validation accuracy 95% 75% Loss Accuracy 0.01 1 30 1 30 12 Epoch 12 Epoch Describe the current status of the model. How do you choose the best model among models saved after each epoch?