Quizmaster

Learning objectives

What you will be able to explain

How is a connectionist neuron described in the lecture?
Match neuron symbols to meanings
Compute total input $h$
Compute outputs of common transfer functions at $h=1$
Which transfer functions are explicitly listed as typical?
What relation between logistic and tanh is highlighted?

Section 01

How is a connectionist neuron described in the lecture? to Compute total input $h$

How is a connectionist neuron described in the lecture?

The slide explicitly states the input-output relationship as a linear filter with a static non-linearity.

Source: 12_connectionist-neurons.pdf, page 5

Match neuron symbols to meanings

All six symbols are defined directly under the connectionist-neuron equation.

Guided checkpoint

Match the notation from the connectionist-neuron equation to the corresponding meaning.

Source: 12_connectionist-neurons.pdf, page 5

Compute total input $h$

h = 0.8\\cdot1 + (-0.4)\\cdot(-2) + 1.2\\cdot0.5 - 0.7 = 0.8 + 0.8 + 0.6 - 0.7 = 1.5

. This follows the lecture equation for total input.

Guided checkpoint

Use $h = \\sum_j w_j x_j - \\theta$ with $\\mathbf{x}=(1,-2,0.5)$ , $\\mathbf{w}=(0.8,-0.4,1.2)$ , $\\theta=0.7$ .

Source: 12_connectionist-neurons.pdf, page 5

Section 02

Compute outputs of common transfer functions at $h=1$ to What relation between logistic and tanh is highlighted?

Compute outputs of common transfer functions at $h=1$

Using the lecture-listed forms:

\\operatorname{ReLU}(h)=\\max(0,h)

\\sigma(h)=\\frac{1}{1+e^{-h}}

, and

\\tanh(h)

. At

h=1

this yields

1

0.73106

, and

0.76159

Guided checkpoint

Evaluate ReLU, logistic, and $\\tanh$ at $h=1$ .

Source: 12_connectionist-neurons.pdf, page 6

Which transfer functions are explicitly listed as typical?

The slide lists ReLU, logistic, and hyperbolic tangent as typical transfer functions.

Guided checkpoint

Select all that apply.

Source: 12_connectionist-neurons.pdf, page 6

What relation between logistic and tanh is highlighted?

The 'Typical transfer functions (2)' slide shows the logistic-to-tanh transformation with scaling/shift.

Source: 12_connectionist-neurons.pdf, page 9

Section 03

Compute $\\sigma(2)$ using the logistic-tanh relationship to Which special neuron is directly linked to classification on the slide?

Compute $\\sigma(2)$ using the logistic-tanh relationship

\\sigma(2)=\\frac{1}{2}(\\tanh(1)+1)=\\frac{1}{2}(0.76159+1)=0.8808

(approx).

Guided checkpoint

Use $\\sigma(h)=\\frac{1}{2}(\\tanh(h/2)+1)$ .

Source: 12_connectionist-neurons.pdf, page 9

Special transfer functions: true or false

All three special-neuron statements are listed on the slide; ReLU belongs to the typical transfer-function list, not the stochastic binary-neuron definition.

Guided checkpoint

Mark each statement as true or false.

Source: 12_connectionist-neurons.pdf, page 11

Which special neuron is directly linked to classification on the slide?

The special-transfer-functions slide labels binary neuron (sign(h)) with classification.

Source: 12_connectionist-neurons.pdf, page 11

Section 04

What does the threshold/bias shortcut notation enable? to Linear neuron output

What does the threshold/bias shortcut notation enable?

The slide introduces shortcut notation for threshold/bias via augmented vectors and compact dot-product form.

Guided checkpoint

Select all correct points implied by the notation slide.

Source: 12_connectionist-neurons.pdf, page 14

In feature extraction, what does $y_{kl}$ denote?

The feature-extraction equation annotates y_kl as feature strength.

Source: 12_connectionist-neurons.pdf, page 16

Linear neuron output

Using

f(h)=\\alpha h

gives

y=1.5\\cdot(-2)=-3

Guided checkpoint

For linear neuron $f(h)=\\alpha h$ with $\\alpha=1.5$ and $h=-2$ , compute $y$ .

Source: 12_connectionist-neurons.pdf, page 11

Section 05

What is w called in the linear-filter feature-extraction slide? to How is extraction/evaluation of features geometrically framed at the end?

What is w called in the linear-filter feature-extraction slide?

The slide states that w describes a feature and is often called receptive field.

Source: 12_connectionist-neurons.pdf, page 16

Which filter type is explicitly shown as an example in this lecture section?

A dedicated slide in section 1.2.2 is titled 'Center-surround filters'.

Source: 12_connectionist-neurons.pdf, page 19

How is extraction/evaluation of features geometrically framed at the end?

The closing slide in the section states partitioning of feature space into two half-spaces.

Source: 12_connectionist-neurons.pdf, page 20

Section 06

Feature extraction: compute two local filter responses to Sort terms: neuron I/O vs feature extraction

Feature extraction: compute two local filter responses

\\mathbf{w}_1\\cdot\\mathbf{x}=1\\cdot5+0\\cdot2+0\\cdot1+(-1)\\cdot4=1

\\mathbf{w}_2\\cdot\\mathbf{x}=0\\cdot5+1\\cdot2+(-1)\\cdot1+2\\cdot4=9

. This directly follows the linear-filter feature-response computation.

Guided checkpoint

Use $y=\\sum_{ij} w_{ij}x_{ij}$ on the same $2\\times2$ patch.

Patch $\\mathbf{x}$
$\\begin{bmatrix}5 & 2 \\\\ 1 & 4\\end{bmatrix}$

Filter	$\\mathbf{w}$
$\\mathbf{w}_1$	$\\begin{bmatrix}1 & 0 \\\\ 0 & -1\\end{bmatrix}$
$\\mathbf{w}_2$	$\\begin{bmatrix}0 & 1 \\\\ -1 & 2\\end{bmatrix}$

Source: 12_connectionist-neurons.pdf, page 16

Match transfer function formulas

ReLU/logistic/tanh are listed in typical transfer functions, and sign(h) is listed in special transfer functions. Sources: 12_connectionist-neurons.pdf, pages 6 and 11.

Guided checkpoint

Match each expression to its function name.

Sort terms: neuron I/O vs feature extraction

The first group is from the connectionist-neuron I/O equation and definitions; the second group is from linear-filter feature extraction. Sources: 12_connectionist-neurons.pdf, pages 5 and 16.

Guided checkpoint

Assign each card to the best matching bucket.

Section 07

What is section 1.2.2 called? to Connectionist neurons and filters: check the statements

What is section 1.2.2 called?

Slide heading for section 1.2.2 is 'Feature Extraction and Evaluation'.

Source: 12_connectionist-neurons.pdf, page 15

Half-space decision value

h=2\\cdot1+(-1)\\cdot3-0=-1

. This scalar decides the side of a linear separator (half-space partitioning). Sources: 12_connectionist-neurons.pdf, pages 5 and 20.

Guided checkpoint

Using $h=\\mathbf{w}^\\top\\mathbf{x}-\\theta$ with $\\mathbf{w}=(2,-1)$ , $\\mathbf{x}=(1,3)$ , $\\theta=0$ , compute $h$ .

Connectionist neurons and filters: check the statements

These follow directly from the neuron equation/definitions and the feature-extraction slides including center-surround examples. Sources: 12_connectionist-neurons.pdf, pages 5, 16, and 19.

Guided checkpoint

Mark each statement true or false.

Section 08

Stochastic binary neuron: flip probability

P(y\rightarrow -y)=\frac{1}{1+\exp(\beta y h)}=\frac{1}{1+\exp(2\cdot1\cdot0.5)}=\frac{1}{1+e^1}\approx0.26894

Guided checkpoint

Use the lecture formula $P(y\rightarrow -y)=\frac{1}{1+\exp(\beta y h)}$ with $\beta=2$ , $y=1$ , and $h=0.5$ .

Compute the flip probability.

Source: 12_connectionist-neurons.pdf, page 11

Knowledge check

Turn understanding into recall.

The quiz now follows the same concepts in scored form. You can return to this lesson from the quiz whenever a gap appears.

Start the quiz →Back to course

What you will be able to explain

How is a connectionist neuron described in the lecture? to Compute total input hhh

How is a connectionist neuron described in the lecture?

Match neuron symbols to meanings

Compute total input hhh

Compute outputs of common transfer functions at h=1h=1h=1 to What relation between logistic and tanh is highlighted?

Compute outputs of common transfer functions at h=1h=1h=1

Which transfer functions are explicitly listed as typical?

What relation between logistic and tanh is highlighted?

Compute sigma(2)\\sigma(2)sigma(2) using the logistic-tanh relationship to Which special neuron is directly linked to classification on the slide?

Compute sigma(2)\\sigma(2)sigma(2) using the logistic-tanh relationship

Special transfer functions: true or false

Which special neuron is directly linked to classification on the slide?

What does the threshold/bias shortcut notation enable? to Linear neuron output

What does the threshold/bias shortcut notation enable?

In feature extraction, what does ykly_{kl}ykl​ denote?

Linear neuron output

What is w called in the linear-filter feature-extraction slide? to How is extraction/evaluation of features geometrically framed at the end?

What is w called in the linear-filter feature-extraction slide?

Which filter type is explicitly shown as an example in this lecture section?

How is extraction/evaluation of features geometrically framed at the end?

Feature extraction: compute two local filter responses to Sort terms: neuron I/O vs feature extraction

Feature extraction: compute two local filter responses

Match transfer function formulas

Sort terms: neuron I/O vs feature extraction

What is section 1.2.2 called? to Connectionist neurons and filters: check the statements

What is section 1.2.2 called?

Half-space decision value

Connectionist neurons and filters: check the statements

Stochastic binary neuron: flip probability

Stochastic binary neuron: flip probability

Turn understanding into recall.

How is a connectionist neuron described in the lecture? to Compute total input $h$

Compute total input $h$

Compute outputs of common transfer functions at $h=1$ to What relation between logistic and tanh is highlighted?

Compute outputs of common transfer functions at $h=1$

Compute $\\sigma(2)$ using the logistic-tanh relationship to Which special neuron is directly linked to classification on the slide?

Compute $\\sigma(2)$ using the logistic-tanh relationship

In feature extraction, what does $y_{kl}$ denote?