7 ms·
In my opinion, the results are not quite exciting as they might seem like at the first glance. The hip-hop and minimal house classification perform almost rando
by oscii 9y ago
In my opinion, the results are not quite exciting as they might seem like at the first glance. The hip-hop and minimal house classification perform almost randomly (the random classifier would have accuracy of 50%). The claim of music genre subjectivity is not fully appropriate for the categories used in this work: the presented genres are quite distinct, and they have objective differences. Knowing only BMP and rhythm structure of the tracks would be sufficient to classify most of the mentioned genres. Also, the article lacks of critical analysis of the results. The network may not have learned to analyze structural properties of the music; if this is true, than what is it classifying exactly? An averaged spectral envelope or spectral distribution? In this case the network will fail if you feed a filtered music piece into it. There is a nice paper on issues like these called “A Simple Method to Determine if a Music Information Retrieval System is a Horse”, you may want to check it out: https://www.researchgate.net/publication/265645782 https://www.researchgate.net/publication/265645782
I understand this is an educational project, but nevertheless it's published, hence open for critics ;)
Edit: small style corrections.
- mattmcknight 9y ago"The hip-hop and minimal house classification perform almost randomly (the random classifier would have accuracy of 50%). " You are assuming that this is a series of binary classifiers. It is multiclass classification, so the base rate for nine classes is 11%.
- oscii 9y agoOh yes, you're right. Thanks for the correction! A recent peer review is still in my head.
- 1r3n1cus 9y agoIf the classes are balanced, that is. Without knowing the distribution of the classes it is difficult to understand if the result is good or not.
- redsparrow 9y agoThe author downloaded 1000 tracks from each class, so they are evenly distributed.
- naushniki 9y agoI agree that the resulting application is rather primitive. Although it was interesting for me, as someone who just learned the theory behind ML, to see how an ML application is built from front to end. I expect that real world applications would encompass a lot of knowledge, which you normally learn after you developed the first version of the app and started using it. I wonder if there are articles out there, which share ordered and filtered information on that.