DM996Autonomous Sensing, Learning and Reasoning
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Session 2022-2023
DM996Autonomous Sensing, Learning and Reasoning
For MSc in Autonomous Robotic Intelligent Systems
Laboratory Task 2:
Optimisation Techniques and Logistic Regression
1 Introduction
This assignment describes the tasks of second laboratory work of the class. In this Computer Lab report, each student should summarise the work undertaken in the Computer Lab exercises.
This Lab Report is an individual assessment and each report account for 10% of the final marks. Submission of this second Lab report will be through MyPlace in week 11 – see link for submission details. Marks and comments will be returned through MyPlace around 3 working weeks later.
The reports will contain the task description, your approach and rationale based on what you learned in the class. The report should also contain your software setup, the data sets, the tools and techniques applied, a brief explanation of the techniques incl. literature references in your own words, and a discussion of the results. Screenshots of code is not enough!
The report will be assessed with respect to
- Structure (format, English, imagery, tidiness)
- Depth of understanding (based on the level of detail and clarity of the explanation)
- Use of references to explain the methods and data sets and the outcomes.
- Critical scholarly discussion (objective and scientifically).
Please avoid using text verbatim (copy and paste) from the literature. Please always use your own words to explain your work.
The length of the report should be 700 words minimum (excluding literature, tables, codes and appendices). There is no maximum word limit to allow you to discuss your topics in the depth deem appropriate. However, try to be concise, within reason and to the point.
You will need to solve the tasks and write the report as a home exercise.
2 Suggested structure of the report
DM996 Autonomous Sensing, Learning and Reasoning – Lab report 2
Name: student name Date: date of submission
Task: In this part you should define briefly the task and your task formulation from assignment task description
Approach: techniques available, technique chosen, brief explanation of the technique and rationale based on literature, requirements and precondition for use of the technique, advantages and disadvantages, expected outcome
Dataset: brief discussion, properties, rationale
Application: code rational and explanation, application of the code, comments in the code
Results: show the output and technical discussion of the result; interpretation of the results (non-technical).
Extras: introduce and explain any observations (outliers, speed, extra work, etc .). Clear and readable figures and screenshots where applicable. The idea of the lab reports is for you to demonstrate that you understood the class contents thoroughly and can apply them in solving given problems.
3 Laboratory report tasks
The lab report covers the following
3.1 Task 1 Optimisation: 2 Marks
Implement a simple hill-climbing and a simple swarm based optimisation algorithm of your choice in Matlab and apply them to these test functions:
a) f(x) = (x-1)2 + 1
b) f(x1 , x2 ) = (x12 + x22 - 2 x1 )2 + 0.25 x1 (Zettl function)
c) f(x1 , x2 ) = 100 (x2 – x12 )2 + (x1 – 1)2 (Rosenbrock function)
Please explain your algorithm and your results.
3.2 Task 2 the Two-bar Truss[1]:
3.2.1 Unconstrained optimisation: 3 Marks
Consider the design of a simple tubular symmetric truss shown inFigure 1. A design of the
truss is specified by a unique set of values for the analysis variables: height (H), diameter, (d), thickness (t), separation distance (B), modulus of elasticity (E), material density (ρ) and load (P). Suppose we are interested in designing a truss that has a minimum weight.
Figure 1 A simple tubular symmetric truss
In this case it is possible can develop a model of the truss using explicit mathematical equation below:
To simplify the design optimisation , let us assume we only consider the height H and separation distance B as design variables, and all the other variables as given as following:
Diameter d (in) |
Thickness t (in) |
Modulus of elasticity E (1000 lbs/in2) |
Density ρ (lbs/in3) |
Load P (1000 lbs) |
3.5 |
0.175 |
30,000 |
0.3 |
66 |
Please implement an optimisation algorithm of your choice to find the best combination of the height and separation distance to minimize the weight.
3.2.2 Constrained optimisation: 1 Mark
You should find that in the unconstrained problem, the optimized values are both near 0 for the height and separation distance, which is impractical. To make the result reasonable, suppose we want to further design the truss that will not yield, will not buckle, and does not deflect "excessively” with the following constraints:
Stress 100
(Stress - Buckling Stress) 0
Deflection 0.25
B > 60
and the model equations:
Could you take these constraints into consideration in your algorithm? For free to apply your own consideration to the constraints.
3.3 Task 3 Heart Disease Prediction: 4 Marks
World Health Organization has estimated 12 million deaths occur worldwide every year due to heart diseases. The early prognosis of cardiovascular diseases can aid in making decisions on lifestyle changes in high risk patients and in turn reduce the complications. A dataset is made available in MY place and is publicly available on the Kaggle website from an ongoing cardiovascular study on residents of the town of Framingham, Massachusetts.
You are tasked to create the 10-year risk of future coronary heart disease (CHD) prediction on this dataset using Logistic Regression and explain your results. Please create a data pipeline choosing Logistic Regression learned in class:
You are allowed to use any software tools you feel most comfortable with, be it MATLAB, Python, R or free or commercial software. Regarding your pipeline, you may either follow the approach demonstrated in the Logistic Regression Laboratory or design your own. However, it is important that in either case, you use your own academic words to explain methods, algorithms, choices/decisions and results. Once you defined the pipeline, you will need to critically analyse your results and you may compare your results with other algorithms at high level and finally state which one you would recommend to use and why.
In your report you can keep the description of the dataset brief and to a minimum to provide just the information you need to explain your approach. Keep your report informative, concise and to the point.
4 Submission
Your report must be uploaded in electronic format to MyPlace by 17:00 pm Wednesday 30th November 2022. Late submission will be penalised by 5% per day unless a valid medical certificate is presented.
2022-12-05