Robotics 2
Machine Learning Unsupervised
Quiz
Your ID Code:
Question 3: In general, which kind of algorithm will work better, supervised or unsupervised?
Question 2: What is the difference between 'supervised' and 'unsupervised' learning algorithms?
Question 1: Every machine learning algorithm consists of what two phases?
Opening and Closing
Supervised and Unsupervised
Training and Testing
Learning and Unlearning
Supervised algorithms require a 'supervisor' who can give the robot a set of rules to follow while solving problems, unsupervised algorithms don't require this.
Supervised algorithms require a 'supervisor' who knows the answers during training, unsupervised algorithms don't require this.
Supervised algorithms can work with data that has patterns that the supervisor doesn't know, while unsupervised algorithms can't do this.
Supervised algorithms can provide 'supervision' for other problem-solving algorithms, while unsupervised algorithms can't do this.
Supervised
Unsupervised
Question 4: Which of these is a reason why we might choose to use an unsupervised, rather than supervised, algorithm?
Sometimes the supervisor makes mistakes in categorizing data. In these cases, unsupervised algorithms will perform better than supervised algorithms.
Question 5: In the k-means clustering algorithm, there is a loop that has to repeat. How do we know when we've repeated the loop enough times?
Questions 6-10: All of the steps of the k-means clustering algorithm are listed here all mixed up. Select a number next to each one to indicate in what order the steps should be done. Each number should only be selected once.
Unsupervised algorithms generally work better than supervised algorithms.
The locations of the 'seeds' no longer change.
Repeating the loop 3 times is always enough.
The locations of the data points to be categorized no longer change.
The supervisor will tell us when the loop has been repeated enough times.
1
2
3
4
5
Calculate the distance between each point in the database and each seed
Define one seed for each category
Each point is assigned to the category of the seed that is closest to it
The new seed is calculated as the center of all the points assigned to its category
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
Loop back to step 2
1
2
3
4
5
Sometimes we (the supervisor) don't know the information we want the robot to learn; we want the robot to find patterns in data that we can't see. Only unsupervised algorithms can be used in these cases.
Supervised algorithms are much more computationally intensive than unsupervised methods. If we need the learning algorithm to run fast, unsupervised algorithms are better.