Comments for FANNFANN

Comment on Advanced Usage by Animesh

Animesh — Thu, 04 Apr 2013 07:21:10 +0000

Is there a way to check the weights of all the hidden neurons at each epoch? or something similar to that

Comment on Getting Started by Xtrim-A

Xtrim-A — Mon, 25 Mar 2013 13:58:48 +0000

Cool new look!!!
I would like to see more examples (different ways of implementation) in documentation page, most possibly by the FANN users who are using it for various applications and using different languages such as c++ and python. Thant will help newbies like me to quickly understand how to use/code FANN more efficiently.

Comment on Updated FANN site by Ben

Ben — Sat, 23 Mar 2013 15:13:27 +0000

Looking good!

Comment on Graphical Interfaces by Mario.

Mario. — Sat, 16 Mar 2013 22:15:54 +0000

Yes it can help in sorting out the traffic problems, but the proper training is very important, you will have to either find, imagine or simulate many example cases.

Also, the whole definition of the problem of the incoming cars and traffic light duration needs to be correctly defined – that is: what the desired result should be – which value are you trying to optimize? Also very important is connection with other traffic lights – the only good system is where all the traffic lights are interconnected and work together as a system! (A good example of synchronisation is that I have been able to drive through city on all green lights, that is about 20 crossings with traffic lights, without stopping.)

Comment on Installing FANN by Bhupendra

Bhupendra — Fri, 15 Mar 2013 01:13:38 +0000

Got it!!!
Posting it for anyone having same issue on Mac.
For OSX you need to change optimization to ’0′ instead of ’3′. Just change these two lines in the ./example/Makefile
>>
%: %.c Makefile
$(GCC) -O0 $< -o $@ -lfann -lm

%_fixed: %.c Makefile
$(GCC) -O0 -DFIXEDFANN $>

Output of the testrun:
bhupendra@vpn examples$ make runtest
gcc -O0 xor_train.c -o xor_train -lfann -lm
gcc -O0 xor_test.c -o xor_test -lfann -lm
gcc -O0 -DFIXEDFANN xor_test.c -o xor_test_fixed -lfixedfann -lm
gcc -O0 simple_train.c -o simple_train -lfann -lm
gcc -O0 steepness_train.c -o steepness_train -lfann -lm
gcc -O0 simple_test.c -o simple_test -lfann -lm
gcc -O0 robot.c -o robot -lfann -lm
gcc -O0 mushroom.c -o mushroom -lfann -lm
gcc -O0 cascade_train.c -o cascade_train -lfann -lm
gcc -O0 scaling_test.c -o scaling_test -lfann -lm
gcc -O0 scaling_train.c -o scaling_train -lfann -lm

Training network
./xor_train
Creating network.
Training network.
Max epochs 1000. Desired error: 0.0000000000.
Epochs 1. Current error: 0.2865722179. Bit fail 4.
Epochs 10. Current error: 0.1378555745. Bit fail 4.
Epochs 20. Current error: 0.0009042433. Bit fail 4.
Epochs 28. Current error: 0.0000391395. Bit fail 0.
Testing network. 0.000023
XOR test (-1.000000,-1.000000) -> -0.991215, should be -1.000000, difference=0.008785
XOR test (-1.000000,1.000000) -> 0.988305, should be 1.000000, difference=0.011695
XOR test (1.000000,-1.000000) -> 0.991032, should be 1.000000, difference=0.008968
XOR test (1.000000,1.000000) -> -0.991266, should be -1.000000, difference=0.008734
Saving network.
Cleaning up.

Testing network with floats
./xor_test
Creating network.
Layer / Neuron 0123456
L 1 / N 3 CCB….
L 1 / N 4 CCC….
L 1 / N 5 BBb….
L 1 / N 6 …….
L 2 / N 7 …BCdd
L 2 / N 8 …….
Input layer : 2 neurons, 1 bias
Hidden layer : 3 neurons, 1 bias
Output layer : 1 neurons
Total neurons and biases : 8
Total connections : 13
Connection rate : 1.000
Network type : FANN_NETTYPE_LAYER
Training algorithm : FANN_TRAIN_RPROP
Training error function : FANN_ERRORFUNC_TANH
Training stop function : FANN_STOPFUNC_BIT
Bit fail limit : 0.010
Learning rate : 0.700
Learning momentum : 0.000
Quickprop decay : -0.000100
Quickprop mu : 1.750
RPROP increase factor : 1.200
RPROP decrease factor : 0.500
RPROP delta min : 0.000
RPROP delta max : 50.000
Cascade output change fraction : 0.010000
Cascade candidate change fraction : 0.010000
Cascade output stagnation epochs : 12
Cascade candidate stagnation epochs : 12
Cascade max output epochs : 150
Cascade min output epochs : 50
Cascade max candidate epochs : 150
Cascade min candidate epochs : 50
Cascade weight multiplier : 0.400
Cascade candidate limit :1000.000
Cascade activation functions[0] : FANN_SIGMOID
Cascade activation functions[1] : FANN_SIGMOID_SYMMETRIC
Cascade activation functions[2] : FANN_GAUSSIAN
Cascade activation functions[3] : FANN_GAUSSIAN_SYMMETRIC
Cascade activation functions[4] : FANN_ELLIOT
Cascade activation functions[5] : FANN_ELLIOT_SYMMETRIC
Cascade activation functions[6] : FANN_SIN_SYMMETRIC
Cascade activation functions[7] : FANN_COS_SYMMETRIC
Cascade activation functions[8] : FANN_SIN
Cascade activation functions[9] : FANN_COS
Cascade activation steepnesses[0] : 0.250
Cascade activation steepnesses[1] : 0.500
Cascade activation steepnesses[2] : 0.750
Cascade activation steepnesses[3] : 1.000
Cascade candidate groups : 2
Cascade no. of candidates : 80
Testing network.
XOR test (-1.000000, -1.000000) -> -0.991215, should be -1.000000, difference=0.008785
XOR test (-1.000000, 1.000000) -> 0.988305, should be 1.000000, difference=0.011695
XOR test (1.000000, -1.000000) -> 0.991032, should be 1.000000, difference=0.008968
XOR test (1.000000, 1.000000) -> -0.991266, should be -1.000000, difference=0.008734
Cleaning up.

Testing network with fixed points
./xor_test_fixed
Creating network.
Layer / Neuron 0123456
L 1 / N 3 CCB….
L 1 / N 4 CCC….
L 1 / N 5 BBb….
L 1 / N 6 …….
L 2 / N 7 …BCdd
L 2 / N 8 …….
Input layer : 2 neurons, 1 bias
Hidden layer : 3 neurons, 1 bias
Output layer : 1 neurons
Total neurons and biases : 8
Total connections : 13
Connection rate : 1.000
Network type : FANN_NETTYPE_LAYER
Decimal point : 13
Multiplier :8192
Bit fail limit : 82
Testing network.
XOR test (-8192, -8192) -> -8192, should be -8192, difference=0.000000
XOR test (-8192, 8192) -> 7888, should be 8192, difference=0.037109
XOR test (8192, -8192) -> 7967, should be 8192, difference=0.027466
XOR test (8192, 8192) -> -8192, should be -8192, difference=0.000000
Cleaning up.
bhupendra@vpn examples$

Comment on Installing FANN by Bhupendra

Bhupendra — Fri, 15 Mar 2013 01:00:09 +0000

Im using FANN on my Macbookpro. I have installed it very easily but testrun is failed. Can someone have a look at it tell me if there is any issue with my installation?
Console output follows:

bhupendra@vpn examples$ make runtest
gcc -O3 xor_train.c -o xor_train -lfann -lm
gcc -O3 xor_test.c -o xor_test -lfann -lm
gcc -O3 -DFIXEDFANN xor_test.c -o xor_test_fixed -lfixedfann -lm
gcc -O3 simple_train.c -o simple_train -lfann -lm
gcc -O3 steepness_train.c -o steepness_train -lfann -lm
gcc -O3 simple_test.c -o simple_test -lfann -lm
gcc -O3 robot.c -o robot -lfann -lm
gcc -O3 mushroom.c -o mushroom -lfann -lm
gcc -O3 cascade_train.c -o cascade_train -lfann -lm
gcc -O3 scaling_test.c -o scaling_test -lfann -lm
gcc -O3 scaling_train.c -o scaling_train -lfann -lm

Training network
./xor_train
Creating network.
Training network.
Max epochs 1000. Desired error: 0.0000000000.
Epochs 1. Current error: 0.2998471856. Bit fail 4.
Epochs 10. Current error: 0.1645403504. Bit fail 4.
Epochs 20. Current error: 0.1343973279. Bit fail 4.
Epochs 30. Current error: 0.1321901083. Bit fail 4.
Epochs 40. Current error: 0.1320076436. Bit fail 4.
Epochs 50. Current error: 0.1318376809. Bit fail 4.
Epochs 60. Current error: 0.1316868365. Bit fail 4.
Epochs 70. Current error: 0.1314437091. Bit fail 4.
Epochs 80. Current error: 0.1314807087. Bit fail 4.
Epochs 90. Current error: 0.1314186156. Bit fail 4.
Epochs 100. Current error: 0.1313483715. Bit fail 4.
Epochs 110. Current error: 0.1314515471. Bit fail 4.
Epochs 120. Current error: 0.1313693523. Bit fail 4.
Epochs 130. Current error: 0.1314102709. Bit fail 4.
Epochs 140. Current error: 0.1313417256. Bit fail 4.
Epochs 150. Current error: 0.1313553154. Bit fail 4.
Epochs 160. Current error: 0.1313396096. Bit fail 4.
Epochs 170. Current error: 0.1313699782. Bit fail 4.
Epochs 180. Current error: 0.1313713491. Bit fail 4.
Epochs 190. Current error: 0.1313034594. Bit fail 4.
Epochs 200. Current error: 0.1313286722. Bit fail 4.
Epochs 210. Current error: 0.1313354671. Bit fail 4.
Epochs 220. Current error: 0.1313285679. Bit fail 4.
Epochs 230. Current error: 0.1312805712. Bit fail 4.
Epochs 240. Current error: 0.1312561035. Bit fail 4.
Epochs 250. Current error: 0.1312738359. Bit fail 4.
Epochs 260. Current error: 0.1312966943. Bit fail 4.
Epochs 270. Current error: 0.1312941313. Bit fail 4.
Epochs 280. Current error: 0.1312871873. Bit fail 4.
Epochs 290. Current error: 0.1312831491. Bit fail 4.
Epochs 300. Current error: 0.1312471628. Bit fail 4.
Epochs 310. Current error: 0.1312719584. Bit fail 4.
Epochs 320. Current error: 0.1312411875. Bit fail 4.
Epochs 330. Current error: 0.1312507540. Bit fail 4.
Epochs 340. Current error: 0.1312562227. Bit fail 4.
Epochs 350. Current error: 0.1312428415. Bit fail 4.
Epochs 360. Current error: 0.1312461942. Bit fail 4.
Epochs 370. Current error: 0.1312564313. Bit fail 4.
Epochs 380. Current error: 0.1312440932. Bit fail 4.
Epochs 390. Current error: 0.1312446594. Bit fail 4.
Epochs 400. Current error: 0.1312422752. Bit fail 4.
Epochs 410. Current error: 0.1312395483. Bit fail 4.
Epochs 420. Current error: 0.1312251985. Bit fail 4.
Epochs 430. Current error: 0.1311910152. Bit fail 4.
Epochs 440. Current error: 0.1311998963. Bit fail 4.
Epochs 450. Current error: 0.1312367022. Bit fail 4.
Epochs 460. Current error: 0.1312312782. Bit fail 4.
Epochs 470. Current error: 0.1312316060. Bit fail 4.
Epochs 480. Current error: 0.1312264204. Bit fail 4.
Epochs 490. Current error: 0.1312348247. Bit fail 4.
Epochs 500. Current error: 0.1311919093. Bit fail 4.
Epochs 510. Current error: 0.1312199533. Bit fail 4.
Epochs 520. Current error: 0.1312156916. Bit fail 4.
Epochs 530. Current error: 0.1312299967. Bit fail 4.
Epochs 540. Current error: 0.1312159747. Bit fail 4.
Epochs 550. Current error: 0.1312028170. Bit fail 4.
Epochs 560. Current error: 0.1311953217. Bit fail 4.
Epochs 570. Current error: 0.1312059760. Bit fail 4.
Epochs 580. Current error: 0.1312075257. Bit fail 4.
Epochs 590. Current error: 0.1312050819. Bit fail 4.
Epochs 600. Current error: 0.1312092394. Bit fail 4.
Epochs 610. Current error: 0.1311953366. Bit fail 4.
Epochs 620. Current error: 0.1312166899. Bit fail 4.
Epochs 630. Current error: 0.1312093884. Bit fail 4.
Epochs 640. Current error: 0.1312118769. Bit fail 4.
Epochs 650. Current error: 0.1312177479. Bit fail 4.
Epochs 660. Current error: 0.1311965585. Bit fail 4.
Epochs 670. Current error: 0.1311824322. Bit fail 4.
Epochs 680. Current error: 0.1311974674. Bit fail 4.
Epochs 690. Current error: 0.1311904341. Bit fail 4.
Epochs 700. Current error: 0.1312078834. Bit fail 4.
Epochs 710. Current error: 0.1312024593. Bit fail 4.
Epochs 720. Current error: 0.1312027872. Bit fail 4.
Epochs 730. Current error: 0.1312017739. Bit fail 4.
Epochs 740. Current error: 0.1312023550. Bit fail 4.
Epochs 750. Current error: 0.1311884224. Bit fail 4.
Epochs 760. Current error: 0.1311852485. Bit fail 4.
Epochs 770. Current error: 0.1312024593. Bit fail 4.
Epochs 780. Current error: 0.1311842203. Bit fail 4.
Epochs 790. Current error: 0.1311897635. Bit fail 4.
Epochs 800. Current error: 0.1311981380. Bit fail 4.
Epochs 810. Current error: 0.1311885417. Bit fail 4.
Epochs 820. Current error: 0.1311869621. Bit fail 4.
Epochs 830. Current error: 0.1311998963. Bit fail 4.
Epochs 840. Current error: 0.1312031895. Bit fail 4.
Epochs 850. Current error: 0.1311789453. Bit fail 4.
Epochs 860. Current error: 0.1311890185. Bit fail 4.
Epochs 870. Current error: 0.1311903894. Bit fail 4.
Epochs 880. Current error: 0.1311752647. Bit fail 4.
Epochs 890. Current error: 0.1311975718. Bit fail 4.
Epochs 900. Current error: 0.1312029064. Bit fail 4.
Epochs 910. Current error: 0.1311938316. Bit fail 4.
Epochs 920. Current error: 0.1311802417. Bit fail 4.
Epochs 930. Current error: 0.1311967373. Bit fail 4.
Epochs 940. Current error: 0.1311729103. Bit fail 4.
Epochs 950. Current error: 0.1311972737. Bit fail 4.
Epochs 960. Current error: 0.1311960965. Bit fail 4.
Epochs 970. Current error: 0.1312002838. Bit fail 4.
Epochs 980. Current error: 0.1312011927. Bit fail 4.
Epochs 990. Current error: 0.1311844289. Bit fail 4.
Epochs 1000. Current error: 0.1312025189. Bit fail 4.
Testing network. 0.131195
XOR test (-1.000000,-1.000000) -> -0.845081, should be -1.000000, difference=0.154919
XOR test (-1.000000,1.000000) -> -0.012242, should be 1.000000, difference=1.012242
XOR test (1.000000,-1.000000) -> 0.843651, should be 1.000000, difference=0.156349
XOR test (1.000000,1.000000) -> 0.012936, should be -1.000000, difference=1.012936
Saving network.
Cleaning up.

Testing network with floats
./xor_test
Creating network.
Layer / Neuron 0123456
L 1 / N 3 Abb….
L 1 / N 4 ACC….
L 1 / N 5 ACC….
L 1 / N 6 …….
L 2 / N 7 …FECa
L 2 / N 8 …….
Input layer : 2 neurons, 1 bias
Hidden layer : 3 neurons, 1 bias
Output layer : 1 neurons
Total neurons and biases : 8
Total connections : 13
Connection rate : 1.000
Network type : FANN_NETTYPE_LAYER
Training algorithm : FANN_TRAIN_RPROP
Training error function : FANN_ERRORFUNC_TANH
Training stop function : FANN_STOPFUNC_BIT
Bit fail limit : 0.010
Learning rate : 0.700
Learning momentum : 0.000
Quickprop decay : -0.000100
Quickprop mu : 1.750
RPROP increase factor : 1.200
RPROP decrease factor : 0.500
RPROP delta min : 0.000
RPROP delta max : 50.000
Cascade output change fraction : 0.010000
Cascade candidate change fraction : 0.010000
Cascade output stagnation epochs : 12
Cascade candidate stagnation epochs : 12
Cascade max output epochs : 150
Cascade min output epochs : 50
Cascade max candidate epochs : 150
Cascade min candidate epochs : 50
Cascade weight multiplier : 0.400
Cascade candidate limit :1000.000
Cascade activation functions[0] : FANN_SIGMOID
Cascade activation functions[1] : FANN_SIGMOID_SYMMETRIC
Cascade activation functions[2] : FANN_GAUSSIAN
Cascade activation functions[3] : FANN_GAUSSIAN_SYMMETRIC
Cascade activation functions[4] : FANN_ELLIOT
Cascade activation functions[5] : FANN_ELLIOT_SYMMETRIC
Cascade activation functions[6] : FANN_SIN_SYMMETRIC
Cascade activation functions[7] : FANN_COS_SYMMETRIC
Cascade activation functions[8] : FANN_SIN
Cascade activation functions[9] : FANN_COS
Cascade activation steepnesses[0] : 0.250
Cascade activation steepnesses[1] : 0.500
Cascade activation steepnesses[2] : 0.750
Cascade activation steepnesses[3] : 1.000
Cascade candidate groups : 2
Cascade no. of candidates : 80
Testing network.
XOR test (-1.000000, -1.000000) -> -0.845081, should be -1.000000, difference=0.154919
XOR test (-1.000000, 1.000000) -> -0.012242, should be 1.000000, difference=1.012242
XOR test (1.000000, -1.000000) -> 0.843651, should be 1.000000, difference=0.156349
XOR test (1.000000, 1.000000) -> 0.012936, should be -1.000000, difference=1.012936
Cleaning up.

Testing network with fixed points
./xor_test_fixed
Creating network.
Layer / Neuron 0123456
L 1 / N 3 Abb….
L 1 / N 4 ACC….
L 1 / N 5 ACC….
L 1 / N 6 …….
L 2 / N 7 …FECa
L 2 / N 8 …….
Input layer : 2 neurons, 1 bias
Hidden layer : 3 neurons, 1 bias
Output layer : 1 neurons
Total neurons and biases : 8
Total connections : 13
Connection rate : 1.000
Network type : FANN_NETTYPE_LAYER
Decimal point : 13
Multiplier :8192
Bit fail limit : 82
Testing network.
XOR test (-8192, -8192) -> -6287, should be -8192, difference=0.232544
Test failed
XOR test (-8192, 8192) -> -122, should be 8192, difference=1.014893
Test failed
XOR test (8192, -8192) -> 6037, should be 8192, difference=0.263062
Test failed
XOR test (8192, 8192) -> -122, should be -8192, difference=0.985107
Test failed
Cleaning up.
make: *** [runtest] Error 255
bhupendra@vpn examples$

Comment on Help by mohamed

mohamed — Mon, 04 Mar 2013 18:31:06 +0000

Hi, I am user of neural network in regression input/output
I have downloaded Fann , but I have some difficulties to compile fann with visual studio, almost of users have this prblem with C code compiling
is possible to obtain help from anyone,
how to train fann with visual studio, how to use it in input/output regression
thank you

Comment on New FANN website by kokting

kokting — Sun, 03 Mar 2013 14:51:51 +0000

Hi, I am currently doing Gesture Recognition and trying to use FANN. Anyone can guide me on how to use this FANN for my project?

Comment on Working with FANN from Java and Scala by Anastasios Tsiolakidis

Anastasios Tsiolakidis — Sat, 23 Feb 2013 22:30:18 +0000

I’ve noticed your link chose to “downgrade” his Java to 32bit for this to work. It looks like that .jar file is “32bit” as well, somehow the whole toolchain has to be 32 bit or 64 bit, but it looked to me like JNA could become arch-agnostic with a couple of “unclean” modifications. Anyway, I am trying to compile FANN for 64 bits with Visual Studio 2012, after converting the project in the laste source.zip it builds the 32 bit version OK, but switching to 64 bits it has problems finding the includes (maybe more, I am stuck at this first error). That is weird to say the least as a VS behaviour, and certainly I am not impressed with the “absence” of automatic x64 builds in VS2012, I vaguely remember it used to be as easy as selecting “Release X64″ instead of the default “Debug”, now this is handled by Configuration and asks a few more questions. Plus, it creates a X64 temp directory during build which could be the reason the include path is borged (all those ..\.. are invalid, perhaps they are not auto-corrected). For the time being I haven’t been able to inspect the Includes in the IDE, which may be caused by imperfect project conversion code. It would be nice to have 64 bit compiles as well by default, after all I did get the x64 Scala as well. Thanks!

Comment on Help by SILVER KWATIRAYO

SILVER KWATIRAYO — Sat, 09 Feb 2013 22:03:43 +0000

As I can see, FANN is for supervized learning. What about unsupervized ones?