Industrial Technology Research Institute

Kai-Jiun Yang

Chi-Tien Sun

Shang-Ho Tsai

Gwo-Giun Lee

A data compression method, a data compression system and an operation method of a deep learning acceleration chip are provided. The data compression method includes the following steps. A filter coefficient tensor matrix of a deep learning model is obtained. A matrix decomposition procedure is performed according to the filter coefficient tensor matrix to obtain a sparse tensor matrix and a transformation matrix, which is an orthonormal matrix. The product of the transformation matrix and the filter coefficient tensor matrix is the sparse tensor matrix. The sparse tensor matrix is compressed. The sparse tensor matrix and the transformation matrix, or the sparse tensor matrix and a restoration matrix, are stored in a memory. A convolution operation result is obtained by the deep learning acceleration chip using the sparse tensor matrix. The convolution operation result is restored by the deep learning acceleration chip using the restoration matrix.

Data compression method, data compression system and operation method of deep learning acceleration chip

A MIMO symbol detection and search method, a decoding circuit and a receiving antenna system are provided. The signal detection and search method includes the following steps. A symbol search tree is obtained, and a plurality of candidate symbols at each layer of the symbol search tree are sorted. The candidate symbols are traversed in sequence at each layer of the symbol search tree. At each layer of the symbol search tree, if a cumulative partial Euclidean distance is greater than or equal to a threshold, un-scanned candidate symbols are excluded. If the cumulative partial Euclidean distance is less than the threshold, the threshold is updated by the cumulative partial Euclidean distance. When all of the candidate symbols have been calculated, an estimated symbol combination is outputted, and the scan of the symbol search tree is terminated.

Growing community of inventors

Zhubei, Taiwan

Kai-Jiun Yang