Can someone recommend a good algorithm for pattern recognition of mixed labels? I’m interested in an algorithm that is robust in identifying individual patterns even when they appear simultaneously. e.g. I have 6 signals, and I train my system to recognize these 6 signals, however, if I signal 1 and 4 appear mixed, I would like the system to predict signals 1 and 4 simultaneously. This is not a “Multi-label classification” problem where a feature set belongs to different labels, e.g. feature “leg” belongs to labels: “humans”, “dogs”, etc. In these problems all the features are known. In my problem, I don’t possess all the features. I only have the features of the individual patterns, but not all the possible mixed features, e.g. I can add signals 1 and 4 or 1 and 8, those would be 2 “new/mixed” sets of features that describe 1+4 and 1+8 respectively. The catch is that I can’t really train the system with all possible combinations because it would be impractical, and to make it more complicated, the mixture is normally non-linear.
I’m trying binary classifiers using LDA and ANN-MLP, e.g. One-Vs-One, One-Vs-All, etc, but surprisingly the best algorithm so far is a single ANN-MLP. It works quite well, but I’m sure someone has developed a more specialized algorithm for this problem. Any suggestions?
Even more specific?
I’m doing patrec of myoelectric signals for the control of prosthetic devices. The typical scenario is placing some bipolar electrodes in the remaining muscles of the stump, each of these electrodes give information of the overall myoelectric activity of its local area. I can easily predict over 10 different hand and wrist movements using this technique, e.g. hand open/close, wrist extension/flexion, etc. Now, what if I need to predict when the user wants to close the hand and flex the wrist simultaneously (mixed-labels). This simultaneous movement generates new myoelectric signal that are obviously composed with those of individual movements in some way (very tricky way). I normally ask the patient to execute only the individual movements, asking him to do all possible combinations is unpractical, so I’m want to avoid this.
On the signals: Myoelectric signal are non-stationary, stochastic with close zero mean. You normally don’t feed directly the signal to the classifier but you do a snapshot or time window, from which you can extract different features such as the RMS value, standard deviation, and so on. These are the features you feed the classifier with.
If anyone would like to try some data, let me know ;)