Professional Documents
Culture Documents
S. Vivekanandan
Cabin: TT 319A
E-Mail: svivekanandan@vit.ac.in
Mobile: 8124274447
Content
HEBB NET
Introduction
Architecture
Linear separability
Algorithm & Flow chart
Examples for Logic function
Examples for Vector function
Features
w XiY
31-01-2017 Dr. S. Vivekanandan Asso. Prof.-SELECT 3
Architecture
w1
Xi Yj
w2
Xn wn
Yin b XiY
b XiY 0
• The relation gives the boundary region of the net input
b x1w1 x 2 w2 0
X 2 X 1W 1 b
W2 W2
• These two regions are called the decision regions of the net.
Start
1
Stop
1
31-01-2017 Dr. S. Vivekanandan Asso. Prof.-SELECT 8
HEBB net for the AND function (Bipolar inputs and targets)
X1 X2 B y
1 1 1 1
1 -1 1 -1
-1 1 1 -1
-1 -1 1 -1
• AND function gives a high ‘1’if both the inputs are high else -1.
• Initially the weights and bias are set to zero
W1 = W 2 = b = 0
The weights change is calculated using
w XiY b y
Wnew Wold W
X1 X2 b Y ΔW1 ΔW2 Δb W1 W2 b
0 0 0
1 1 1 1
1 -1 1 -1
-1 1 1 -1
-1 -1 1 -1
X1 X2 b Y ΔW1 ΔW2 Δb W1 W2 b
0 0 0
1 1 1 1 1 1 1 1 1 1
1 -1 1 -1 -1 1 -1 0 2 0
-1 1 1 -1 1 -1 -1 1 1 -1
-1 -1 1 -1 1 1 -1 2 2 -2
X1 X2 b Y ΔW1 ΔW2 Δb W1 W2 b
0 0 0
1 1 1 1 1 1 1 1 1 1
1 -1 1 -1 -1 1 -1 0 2 0
-1 1 1 -1 1 -1 -1 1 1 -1
-1 -1 1 -1 1 1 -1 2 2 -2
This completes one epoch of training. The straight line can be obtained by
X 2 X 1W 1 b
W2 W2
epoch W1 W2 b X2
fisrt 1 1 1 -X1 - 1
second 0 2 0 0
third 1 1 -1 -X1+1
forth 2 2 -2 -X1+1
31-01-2017 Dr. S. Vivekanandan Asso. Prof.-SELECT 12
HEBB net for AND
1
(-1, 1) (1, 1)
X1 -2
2
(-1,-1) (1,-1)
X2 Yj
2
X1 X2 B y
1 1 1 1
1 0 1 0
0 1 1 0
0 0 1 0
• AND function gives a high ‘1’if both the inputs are high else -1.
• Initially the weights and bias are set to zero
W1 = W 2 = b = 0
The weights change is calculated using
w XiY b y
X1 X2 b Y ΔW1 ΔW2 Δb W1 W2 b
0 0 0
1 1 1 1 1 1 1 1 1 1
1 0 1 0 0 0 0 1 1 1
0 1 1 0 0 0 0 1 1 1
0 0 1 0 0 0 0 1 1 1
This completes one epoch of training. The straight line can be obtained by
X 2 X 1W 1 b
W2 W2
epoch W1 W2 b X2
fisrt 1 1 1 -X1 - 1
second 1 1 1 -X1 - 1
third 1 1 1 -X1-1
forth 1 1 1 -X1-1
31-01-2017 Dr. S. Vivekanandan Asso. Prof.-SELECT 15
HEBB net for the AND function (Binary inputs and Bipolar targets)
X1 X2 B y
1 1 1 1
1 0 1 -1
0 1 1 -1
0 0 1 -1
• AND function gives a high ‘1’if both the inputs are high else -1.
• Initially the weights and bias are set to zero
W1 = W 2 = b = 0
The weights change is calculated using
w XiY b y
X1 X2 b Y ΔW1 ΔW2 Δb W1 W2 b
0 0 0
1 1 1 1 1 1 1 1 1 1
1 0 1 -1 -1 0 -1 0 1 0
0 1 1 -1 0 -1 -1 0 0 -1
0 0 1 -1 0 0 -1 0 0 -2
This completes one epoch of training. The straight line can be obtained by
X 2 X 1W 1 b
W2 W2
epoch W1 W2 b X2
fisrt 1 1 1 -x1-1
second 0 1 0 0
third 0 0 -1 0
forth 0 0 -2 0
31-01-2017 Dr. S. Vivekanandan Asso. Prof.-SELECT 17
• This can be continued with OR, XOR, NAND, ANDNOT, etc…..
X1 X2 X3 X4 b y
1 1 1 1 1 1
1 1 1 -1 1 -1
-1 1 -1 -1 1 1
1 -1 -1 1 1 -1
EEE 408-NNFL-NRB-VIT
General learning rule – Vector based problems
EEE 408-NNFL-NRB-VIT
Hebbian Learning for Vector Problems
EEE 408-NNFL-NRB-VIT
Rule: When an axon of cell A is near enough to excite a cell B and
repeatedly or persistently takes place in firing it, some growth
process or metabolic change takes place in one or both cells such
that A's efficiency, as one of the cells firing B, is increased.
Example:
needs to be trained using the set of three input vectors as below: