Description
Description:
1. Random Data Generator
a. Univariate gaussian data generator
Input
Expectation value or mean:
Variance:
Output: A data point from
HINT
Generating values from normal distribution
You have to handcraft your geneartor based on one of the approaches given in the
hyperlink.
You can use uniform distribution function (Numpy)
b. Polynomial basis linear model data generator
is a
Input:
vector
(basis number),
e.g.
Output:
,
(a number)
Internal constraint
is uniformly distributed.
2. Sequential Estimator
Sequential estimate the mean and variance
Data is given from the univariate gaussian data generator (1.a).
Input:
as in (1.a)
Function:
Call (1.a) to get a new data point from
Use sequential estimation to find the current estimates to
andRepeat steps above until the estimates converge.
Output: Print the new data point and the current estimiates of
and
in each iteration.
Notes
You should derive the recursive function of mean and variance based on the sequential
esitmation.
Hint: Online algorithm
Sample input & output (
1
for reference only
)
Data point source function: N(3.0, 5.0)
2
3 Add data point: 3.234685454257290
4 Mean = 3.408993960833291
5 Add data point: 0.519242879651157
6 Mean = 2.445743600439247
7 Add data point: 1.347113997201991
8 Mean = 2.171086199629932
9 Add data point: 8.979491998496083
Variance = 0.030383455464755956
Variance = 1.875958150575018
Variance = 1.633278676389248
10 Mean = 3.532767359403163
Variance = 8.723325264636875
11 Add data point: 3.603448448693051
12 Mean = 3.544547540951477
13 Add data point: 4.127197937610908
14 Mean = 3.627783311902824
15 Add data point: 4.992735798186870
16 Mean = 3.798402372688330
Variance = 7.270131583917285
Variance = 6.273110519038578
Variance = 5.692747751482052
17
18
…
19
20 Add data point: 4.233592159021013
21 Mean = 2.961576104513964
22 Add data point: 3.529990930040463
23 Mean = 2.961883688294010
24 Add data point: 1.125210345431449
25 Mean = 2.960890354955524
Variance = 5.045715437349161
Variance = 5.043159812425648
Variance = 5.042255747918937
3. Baysian Linear regression
Input
The precision (i.e., b) for initial prior
All other required inputs for the polynomial basis linear model geneartor (1.b)
Function
Call (1.b) to generate one data point
Update the prior, and calculate the parameters of predictive distribution
Repeat steps above until the posterior probability converges.
Output
Print the new data point and the current paramters for posterior and predictive
distribution.After probability converged, do the visualization
Ground truth function (from linear model generator)
Final predict result
At the time that have seen 10 data points
At the time that have seen 50 data points
Note
Except ground truth, you have to draw those data points which you have seen
before
Draw a black line to represent the mean of function at each point
Draw two red lines to represent the variance of function at each point
In other words, distance between red line and mean is ONE variance
Hint: Online learning
Sample input & output (for reference only)
1. b = 1, n = 4, a = 1, w = [1, 2, 3, 4]
1
Add data point (-0.64152, 0.19039):
2
3 Postirior mean:
4 0.0718294547
5 -0.0460797888
6 0.0295609502
7 -0.0189638408
8
9
Posterior variance:
10 0.6227289276, 0.2420256620, -0.1552634839, 0.0996041049
11 0.2420256620, 0.8447365161, 0.0996041049, -0.0638976884
12 -0.1552634839, 0.0996041049, 0.9361023116, 0.0409914289
13 0.0996041049, -0.0638976884, 0.0409914289, 0.9737033172
14
15 Predictive distribution ~ N(0.00000, 2.65061)
16 ————————————————–
17 Add data point (0.07122, 1.63175):
18
19 Postirior mean:
20 0.6736864869
21 0.2388980107
22 -0.1054659080
23 0.0710615952
24
25 Posterior variance: 26 0.3765992302, 0.1254838660, -0.1000441911, 0.0627881634
27 0.1254838660, 0.7895542671, 0.1257503020, -0.0813299447
28 -0.1000441911, 0.1257503020, 0.9237138418, 0.0492510997
29 0.0627881634, -0.0813299447, 0.0492510997, 0.968196409430
31 Predictive distribution ~ N(0.06869, 1.66008)
32 ————————————————–
33 Add data point (-0.19330, 0.24507):
34
35 Postirior mean:
36 0.5760972313
37 0.2450231522
38 -0.0801842453
39 0.0504992402
40
41 Posterior variance: 42 0.2867129751, 0.1311255325, -0.0767580827, 0.0438488542
43 0.1311255325, 0.7892001707, 0.1242887609, -0.0801412282
44 -0.0767580827, 0.1242887609, 0.9176812972, 0.0541575540
45 0.0438488542, -0.0801412282, 0.0541575540, 0.9642058389
46
47 Predictive distribution ~ N(0.62305, 1.34848)
48 ————————————————–
49
50
…
51
52 ————————————————–
53 Add data point (-0.76990, -0.34768):
54
55 Postirior mean:
56 0.9107496675
57 1.9265499885
58 3.1119297129
59 4.1312375189
60
61 Posterior variance: 62 0.0051883836, -0.0004416700, -0.0086000319, 0.0008247001
63 -0.0004416700, 0.0401966605, 0.0012708906, -0.0554822477
64 -0.0086000319, 0.0012708906, 0.0265353911, -0.0031205875
65 0.0008247001, -0.0554822477, -0.0031205875, 0.0937197255
66
67 Predictive distribution ~ N(-0.61566, 1.00921)
68 ————————————————–
69 Add data point (0.36500, 2.22705):
70
71 Postirior mean:
72 0.9107404583
73 1.9265225090
74 3.1119408740
75 4.1312734131
76
77 Posterior variance:
78 0.0051731092,
-0.0004872471,
-0.0085815201,
0.000884234079 -0.0004872471, 0.0400606628, 0.0013261280, -0.0553046044
80 -0.0085815201, 0.0013261280, 0.0265129556, -0.0031927398
81 0.0008842340, -0.0553046044, -0.0031927398, 0.0934876838
82
83 Predictive distribution ~ N(2.22942, 1.00682)
