Docsity
Docsity

Prepare for your exams
Prepare for your exams

Study with the several resources on Docsity


Earn points to download
Earn points to download

Earn points by helping other students or get them with a premium plan


Guidelines and tips
Guidelines and tips

Layer Perceptron - Introduction to Artificial Intelligence - Solved Exams, Exams of Artificial Intelligence

Main points of this past exam are: Layer Perceptron, Expansion, Collision, Desired Location, Current Position, Optical Flow, Flow Field, Information, Vehicles, Layer Perceptron

Typology: Exams

2012/2013

Uploaded on 04/02/2013

shalin_p01ic
shalin_p01ic 🇮🇳

4

(7)

86 documents

1 / 3

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
CS 188 Midterm Exam, 2-330 pm, Mon Oct 19, 1998
1. (a) If you consider the flow field, then the FOE is the point where all the lines con-
taining flow vectors intersect. This is the point that appears stationary.
To get the time to collision, recall that if you consider the frame of reference in
the image plane that has been shifted so that the FOE is at the origin, then the
flow vector at the image point (with coordinates measured relative
to FOE) is given by
where is the distance from the target to the camera, is the normal-to-the-
image-plane component of the camera’s velocity with respect to the target. Notice
that , and so . Hence the Time-to-collision can be
found as the ratio of an image point’s distance to the FOE to the point’s apparent
speed.
(b) To go towards a target, make sure the target appears stationary (i.e., is at the FOE).
The time to collision shouldtell you when to break.
(c) Only if there is an object you know to be100 feet away, so you can steer towards
it. Optical flow (or monocular images in general) does not allow you to measure
absolute distances.
2. (a) The length should satisfy the condition:
(where and are the prior probabilities for each vehicle type).
With a few algebraic manipulations, you get:
from which you get
1
pf3

Partial preview of the text

Download Layer Perceptron - Introduction to Artificial Intelligence - Solved Exams and more Exams Artificial Intelligence in PDF only on Docsity!

CS 188 Midterm Exam, 2-330 pm, Mon Oct 19, 1998

  1. (a) If you consider the flow field, then the FOE is the point where all the lines con- taining flow vectors intersect. This is the point that appears stationary. To get the time to collision, recall that if you consider the frame of reference in the image plane that has been shifted so that the FOE is at the origin, then the flow vector at the image point (with coordinates measured relative to FOE) is given by

where is the distance from the target to the camera, is the normal-to-the- image-plane component of the camera’s velocity with respect to the target. Notice that , and so. Hence the Time-to-collision can be found as the ratio of an image point’s distance to the FOE to the point’s apparent speed. (b) To go towards a target, make sure the target appears stationary (i.e., is at the FOE). The time to collision should tell you when to break. (c) Only if there is an object you know to be 100 feet away, so you can steer towards it. Optical flow (or monocular images in general) does not allow you to measure absolute distances.

  1. (a) The length should satisfy the condition:

(where and are the prior probabilities for each vehicle type). With a few algebraic manipulations, you get:

from which you get

(b) The perceptron will have two weights: one the multiplicative weight for the length, the other the threshold. The output of the perceptron will be

(c) Let’s define

Using the form of given above, and the error

(where is the desired output), we get

and, similarly,

Thus, for the 8ft truck, assuming the desired output of 1 for trucks and 0 for cars, we perform the updates

  1. Here is the belief net:

S

L P

X

The probabilities you will need are: , and. You can find them by computing the proportion of people who smoke, occurence rates of lung cancer and poor stamina among people who smoke and among those who don’t, and the proportion of spots found in X-ray images of people with cancer and those without.