feature_addition : Added a mean squared loss layer to DNN with test case

Added mean squared loss layer "loss_mean_squared" to
DNN as requested in https://github.com/davisking/dlib/issues/152
Also added test case of a simple linear regression with one variable
that uses this layer.

unicode: Disable misleading-indentation warnings with GCC pragma

increasing loss more robust.  Now it will never get into a situation where it
backtracks over and over.  Instead, it will only backtrack a few times in a row
before just letting SGD run unimpeded.

reachable.

Remove initializer lists support for Apple/Clang (libstdc++ version used does not include std::initializer_list) (#334)

Remove initializer lists support for Apple/Clang (libstdc++ version used does not include std::initializer_list)

* Expose python binding for running multiple detectors at once

* Remove unwanted typedef

Modify to support both train-images-idx3-ubyte and train-images.idx3-ubyte styles of the filename. (#327)

* Add option --rm-other-labels to imglab

* Add support for comma separated list of labels

regularization to decay away is activated.

character, since that is interpreted by matlab as part of an eventual printf()
code.

is built as a stand alone library and what appears in the config.h file.

training_label_type instead of a single label_type.  This way, the label
type used for training can be distinct from the type output by the network.
This change breaks backwards compatibility with the previous API.

this frustrates Visual Studio users who want to compile dlib as a static
library.

versions were calling into cuDNN, however, the cuDNN functions for doing this
are horrifically slow, well over 100x slower than they should be, which is
surprising since these functions are so trivial.