Deep learning is a unsupervised feature extraction algorithm. The training process of it usually contains large number matrix operations. It can be quite time-consuming as data size increases. Intel Xeon Phi is a many-core platform introduced by Intel which is suitable for vector operations. With the help Intel MKL math lib, The platform can deal with the training process easily.

• Parallelize Deep Learning algorithm to enable it to deal with big data.
• Tap into the computational potential of the newly-introduced many-core platform--Intel Xeon Phi.
• Probe the practiability of using Intel Xeon Phi as a new computational platform other than GPU.

For more information about the design of Dolphin and up-to-date documentation on many of our research ideas, check out our website：https://github.com/PasaLab/dolphin.