GSIT - GSI Technology's Product Could Revolutionise Deep Learning

• Search and recommendation processes are already very important in how much mass-adopted technology functions.
• Performance on sparse matrix operations could go beyond these already large markets and affect the whole field.
• Neurons that fire together, wire together, and models represented by sparse matrices represent this natural phenomenon.
• APUs could open up how many applications of deep learning can benefit from sparse matrices that increase the ability of complex models to generalise.
• Search and recommendation is only the beginning for GSI Technologies if they can get the APIs ready for when the product actually starts flowing through foundries.

For further details see: