In the following comparison tables, I have included around twenty libraries for machine learning. I considered such characteristics as the language of implementation and interface, the availability and type of acceleration, license type, ongoing development status, and compatibility with popular package managers. Later in this chapter, we will look at the unique features of each library in more detail.
Table 2.1: Comparison of general-purpose machine learning libraries for iOS (part 1):
|
Library |
Language |
Algorithms |
|
AIToolbox |
Swift |
LinReg, LogReg, GMM, MDP, SVM, NN, PCA, k-means, genetic algorithms, DL: LSTM, CNN. |
|
BrainCore |
Swift |
DL: FF, LSTM. |
|
Caffe, Caffe2, MXNet, TensorFlow, tiny-dnn |
C++ |
DL. |
|
dlib |
C++ |
Bayesian networks, SVMs, regressions, structured prediction, DL, clustering and other unsupervised... |
