16:59-17:27=0.28-> 27
17:42-18:05=0.23-> 22/ 49


IF a_1< x_1< a_1+delta 

(9) GA approach -1
((x_1)(rule-1),(x_2)(rule-2),..., (x_n)(rule-n)) 

(9) GA approach -2

(9) exhaustive


(9) An Immune approach - clonal selection
This algorithm is proposed by ... for the purpose of pattern classification being based on the algorithm called CLONALG proposed by ... for the purpose of pattern recognition.


0. Prepare a set of antigenic patterns Ag.
1. Randomly generate an initial population of antibodies Ab. This is made up of two population antibody memory Abm and antbody repertoir Abr  (Abm + AbR = Ab)
3. Select an antigen Ag(i) from the population Ag.
4. For i=1 to N
       For j=1 to M
         calculate affinity of Ab(j) to Ag(i) 
5. Select the n highest affinity antibodies and generate a number of clones for each
antibody in proportion to their affinity.
6. Mutate these close one by one with an inversely proportional to their affinity
and place them in a newe population of  mature clones m.Clones
7. Calculate affinity of all members of m_Clones to each of antigens Ag(i) (i=1,2,...,N) and select the highest socre one as candidate memory, compare the current memory and if the candidate affinity is greater than the current memory then replace current with candidate, then the candidate becomes the new memory cell.	 
8. Sort Ab_r according to affinity (the higher the upper) and replace the antibodies from the top one by one with the clones from m_Clones.
9. Remove the rest of antibodies (low affinity ones) in the population Abr and replace
with new randomly generated members.
5. Return to step 3 until all antigens have been presented.


memo for the above
==================================

CLONAL G
3 populations
Antibody-memory   $Ab_m$    m=1,2, ..., M
Antibody-reservoire  $Ab_r$  r=1,2, ..., R
Antigen  $Ag_i$  i=1,2,..., N

affinity, e.g., as Hamming distance

3. Select an antigen Ag_i from the population Ag.    
4. For i=1 to N
       For j=1 to M
         calculate affinity of Ag_i to  Ab_j as Hamming distance
5. Select the n highest affinity antibodies and generate a number of clones for each
antibody in proportion to their affinity, placing the clones in a new population Cl_i

6. Mutate the clone Cl_i with an inversely proportional to their affinity
to produce a mature population Cl_i*

7. Calculate affinity of Cl_i* to each of Ag_i (i=1,2,...,N) and select the highest socre one as candidate memory,
compare the current memory, If the candidate affinity is greater than the current memory then replace current with candidate.
then the candidate becomes the new memory cell.e new memory cell.		 

8. Remove those antibodies with low affinity in the population Ab_r and replace them
with new randomly generated members.

9. Repeat steps 3-8 until all antigens have been presented. This represents one generation
of the algorithm.


Note that
This algorithm
retains only one memory cell for each antigen presented to it





1. Select an antigen from the population to be classified
2. Compare the antigen with each memory cell
3. Calculate the percentage of accuracy (affinity/max)*100
4. If percentage is > current highest, make the new candidate the highest
5. If highest is > the threshold level, set classification to memory cell highest
6. Add the result to the set of classifications
7. Loop until all antigens have been presented