Architectural Support for Emerging Neural Networks
The recent rise of deep learning applications has incurred significant computations, which the conventional hardware cannot handle. To meet the computational demands, we need to design specialized domain-specific architectures. In this talk, I will introduce one of the emerging neural networks, Memory-augmented Neural Network (MANN), and discuss its scalability problems. Then, I will present MnnFast, a fast and scalable system architecture for MANNs.
Hanhwi Jang is an assistant professor in the Department of Electrical and Computer Engineering at Ajou University. His research interests lie in the broad area of computer architecture with an emphasis on performance modeling and optimization. Before joining Ajou University, he was a staff
engineer at Samsung Research, where he performed research on deep learning accelerators. He received his PhD and BSc degrees in Computer Science and Engineering from POSTECH.
ZOOM : https://zoom.us/j/8978217407?pwd=d1pOZmF1OWlseEdZRVBpV3VuSkl3dz09
ID : 897 821 7407
PW : 1nTQDY