Summary
- Plugin name : Epoch variance
- Version : 1.0
- Author : Dieter Devlaminck
- Company : INRIA
- Short description : Computes variance of each sample over several epochs
- Documentation template generation date : Dec 3 2018
Description
This box is an extension of the Epoch Average box. It offers several methods of averaging for epoched streams but also outputs variance and confidence bounds.
Inputs
The input type of this box can be changed. Its type must be derived of type Doc_Streams_StreamedMatrix in order to be parsed by the input reader. If the author changes the input type, the output type will be changed the same way.
1. Input epochs
This input receives the input streamed matrix to average.
- Type identifier : Streamed matrix (0x544a003e, 0x6dcba5f6)
Outputs
The output type of this box can be changed. Its type must be derived of type Doc_Streams_StreamedMatrix in order for the writer to format the output chunks. If the author changes the output type, the input type will be changed the same way.
1. Averaged epochs
This output sends the averaged streamed matrix. Averaging method is done according to the box settings.
- Type identifier : Streamed matrix (0x544a003e, 0x6dcba5f6)
2. Variance of epochs
This output sends the variance of the input. Averaging method is done according to the box settings.
- Type identifier : Streamed matrix (0x544a003e, 0x6dcba5f6)
3. Confidence bounds
This output sends the confidence bounds of the input. Averaging method is done according to the box settings.
- Type identifier : Streamed matrix (0x544a003e, 0x6dcba5f6)
Settings
1. Averaging type
This setting gives the method to use in order to average the input matrices. It can be of two types :
- Moving average : in this case, the averaging is done at every input reception on the last few buffers, starting as soon as enough input has been received.
- Moving average (Immediate) : in this case, the averaging is done at every input reception on the last few buffers, starting immediately. When the number of received buffer is lower than the wished number of epochs, the average is computed on this very few number of input buffers.
- Epoch block average : in this case, the averaging is done on a number of epochs (see next setting). Once this exact number of input is received, the average is computed and output.
- Cumulative average : in this case, the averaging is done on an infinite number of epochs starting from the first received buffer to the last received buffer. This can be very memory consuming !
Type identifier : Epoch Average method (0x6530bdb1, 0xd057bbfe)
- Default value : [ Moving epoch average ]
2. Epoch count
This setting tells the box how much buffer it should use in order to compute the average.
- Type identifier : Integer (0x007deef9, 0x2f3e95c6)
- Default value : [ 4 ]
3. Significance level
Significance Level for the confidence bound computation. The higher it is, the tighter the confidence interval will be.
- Type identifier : Float (0x512a166f, 0x5c3ef83f)
- Default value : [ 0.01 ]
Examples
Let's study two cases. First, suppose you have such box with Epoch block average set and four epochs. The input stream is as follows :
+----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ | I1 | | I2 | | I3 | | I4 | | I5 | | I6 | | I7 | | I8 | | I9 | ... +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+
The output stream will look like this :
+----+ +----+ | O1 | | O2 | ... +----+ +----+
where O1
is the average of I1
, I2
, I3
and I4
and where O2
is the average of I5
, I6
, I7
and I8
.
Now consider the case where you configured this box with Moving average and four epochs. Given the same input stream :
+----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ | I1 | | I2 | | I3 | | I4 | | I5 | | I6 | | I7 | | I8 | | I9 | ... +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+
The output stream will look like this :
+----+ +----+ +----+ +----+ +----+ +----+ | O1 | | O2 | | O3 | | O4 | | O5 | | O6 | ... +----+ +----+ +----+ +----+ +----+ +----+
where :
O1
is the average ofI1
,I2
,I3
andI4
O2
is the average ofI2
,I3
,I4
andI5
O3
is the average ofI3
,I4
,I5
andI6
O4
is the average ofI4
,I5
,I6
andI7
- etc...
Again consider the case where you configured this box with Moving average (Immediate) and four epochs. Given the same input stream :
+----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ | I1 | | I2 | | I3 | | I4 | | I5 | | I6 | | I7 | | I8 | | I9 | ... +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+
The output stream will look like this :
+----+ +----+ +----+ +----+ +----+ +----+ | O1 | | O2 | | O3 | | O4 | | O5 | | O6 | ... +----+ +----+ +----+ +----+ +----+ +----+
where :
O1
is exactlyI1
O2
is the average ofI1
andI2
O3
is the average ofI1
,I2
andI3
O4
is the average ofI1
,I2
,I3
andI4
O5
is the average ofI2
,I3
,I4
andI5
O6
is the average ofI3
,I4
,I5
andI6
- etc...
Finally consider the case where you configured this box with Cumulative average and four epochs. Given the same input stream :
+----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ | I1 | | I2 | | I3 | | I4 | | I5 | | I6 | | I7 | | I8 | | I9 | ... +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+ +----+
The output stream will look like this :
+----+ +----+ +----+ +----+ +----+ +----+ | O1 | | O2 | | O3 | | O4 | | O5 | | O6 | ... +----+ +----+ +----+ +----+ +----+ +----+
where :
O1
is exactlyI1
O2
is the average ofI1
andI2
O3
is the average ofI1
,I2
andI3
O4
is the average ofI1
,I2
,I3
andI4
O5
is the average ofI1
,I2
,I3
,I4
, andI5
O6
is the average ofI1
,I2
,I3
,I4
,I5
, andI6
- etc...
Miscellaneous
Generated on Tue Jun 26 2012 15:25:54 for Documentation by 1.7.4