923 reputation
216
bio website graphics.tu-bs.de
location Germany
age 32
visits member for 1 year, 10 months
seen 8 hours ago

Oct
17
comment Profiling from Mathematica
Unfortunately this solution does not seem to work for more complicated constructs, like funcitons called within Module, etc..
Oct
13
comment Using Predict on images
@ChipHurst: I think you have a point there. I first neglegted Classify because I would basically assign a different value to each image. But the problem itself can be reformulated so that I would just say the current image I am looking at has a certain probability to lie in class A or some other class. So I could take that information and assign an appropriate number to the image not in the training data set. To do so I would of course create classes of similar images on the training data set, which often do not contain many images. Results so far are promising.
Oct
13
comment Keyboard shortcut for changing style of selected text (NOT the whole cell)
If you want to change the hotkeys you need to edit "KeyEventTranslations.tr" or for Menu entries the "MenuSetup.tr" file. There are quite some resources on the web and these forums on how to do that. The location of those files is OS dependent, of course.
Oct
10
comment Using Predict on images
@Mr.Wizard: Thanks, the problem does not seem to be dependent on my image dataset, so your suggestion works fine to present the problem. I edited the post, so everyone can try on their own now.
Oct
10
revised Using Predict on images
added 248 characters in body
Oct
10
asked Using Predict on images
Oct
9
comment How to evaluate complex numerical integral in mathematica?
The issue here is not that your integrand is complex, it is the definition of the integrand itself. You might want to try numerical evaluation. For that you could use the pattern _?NumericQ for your integrand parameters.
Sep
24
comment How to get the centroids of the ClusterComponents function
I believe the author of the thread meant the actual centers of the clustering process. Those do not correspond to the mean value of the position vectors. Especially not with considering that the DistanceFunction for the KMeans clustering process has been set to CorrelationDistance.
Sep
24
comment Compile nested Sums
For me, Plus@@Table is slower than Sum inside a Compile statement, but Total@Table is faster, so this kind of answers my question. I always thought building a table first has to be slower, but it seems to have little influence in this case. Thanks.
Sep
24
revised Compile nested Sums
editing to avoid confusion about what problem to solve
Sep
24
comment Compile nested Sums
@xzczd: Thanks for the comment, but the problem above is not the problem I want to solve, it's just an example. The introduction of data seems to cause even more confusion, so I will edit it out again. Basically what I am trying to ask is how to compile nested Sums efficiently.
Sep
24
revised Compile nested Sums
changed code examples
Sep
24
comment Compile nested Sums
@b.gatessucks: This is of course not my real problem, but the real problem has the same structure. I edited the example slightly to show that there is a dependency on external data which leaves no analytic solution.
Sep
24
asked Compile nested Sums
Sep
22
comment Bug in ClusteringComponents yielding different results on practically same input?
Thanks, I am actually feeling a little bit dumb, that I did not think of that myself. Can anyone think of a similar method that does not run into local optimums (I know that RandomSeed helps, but it has still a random component in it)?
Sep
21
asked Bug in ClusteringComponents yielding different results on practically same input?
Aug
25
comment Listable compiled functions used in NMinimize
Ok, I have not tested this, but in theory this should be possible without any additional hassle. At least when using certain optimization algorithms that only rely on evaluating the objective function directly.
Aug
25
comment Using one complicated equation to reduce another
Take a look at Replace. It might be what you are looking for if you actually have a form that already contains the y_j directly. After replacing you might want to apply Simplify or FullSimplify.
Aug
23
accepted Export Grayscale Bitmap
Aug
22
comment Export Grayscale Bitmap
Ok, definitely not a version 10 problem. Same behavior here on version 9 and I think the answers kind of explain it.