Instructions
Answer the proposed question about the iterative approach for a one-parameter predictive model already viewed in lecture.
Show your calculations clearly and briefly explain your reasoning.
This activity is worth 20 points toward your final grade for this unit.
To receive the full 20 points, you must:
- Complete all requested calculations.
- Explain the parameter update process in your own words.
- Submit your answer before Friday, 4th of September.
How to estimate the parameters of a temperature sensor?
A simple temperature sensor is modeled as:

where x is temperature in °C, y is the predicted sensor voltage in mV, and theta is an unknown sensitivity parameter in mV/°C.
The model initially assumes

During an experiment, the following training instance is measured:
x=40°C and y=240 mV
Using the iterative learning approach:
-
Calculate the predicted output
y^using the current value oftheta^0. -
Calculate the residual error:

-
Determine the required correction:

-
Update the model using a learning rate of
alpha=0.2:

- Calculate the new prediction for
x=40usingtheta^1. - Interpret the result: Did the new prediction move closer to the measured value? Explain why the learning rate prevents the model from immediately adapting completely to a single measurement.
- What could happen if one temperature measurement were an outlier and the model used (\alpha=1) for every update?
