Abstract: Systems, methods, and devices for pruning a convolutional neural network (CNN). A subset of layers of the CNN is chosen, and for each layer of the subset of layers, how salient each filter in the layer is to an output of the CNN is determined, a subset of the filters in the layer is determined based on the salience of each filter in the layer, and the subset of filters in the layer is pruned. In some implementations, the layers of the subset of layers of the CNN are non-contiguous. In some implementations, the subset of layers includes odd numbered layers of the CNN and excludes even numbered layers of the CNN. In some implementations, the subset of layers includes even numbered layers of the CNN and excludes odd numbered layers of the CNN.

Abstract: Systems, methods, and devices for pruning a convolutional neural network (CNN). A subset of layers of the CNN is chosen, and for each layer of the subset of layers, how salient each filter in the layer is to an output of the CNN is determined, a subset of the filters in the layer is determined based on the salience of each filter in the layer, and the subset of filters in the layer is pruned. In some implementations, the layers of the subset of layers of the CNN are non-contiguous. In some implementations, the subset of layers includes odd numbered layers of the CNN and excludes even numbered layers of the CNN. In some implementations, the subset of layers includes even numbered layers of the CNN and excludes odd numbered layers of the CNN.

Abstract: The present disclosure disclose method and apparatus for determining memory requirement for processing a DNN model on a device, a method includes receiving a DNN model for an input, wherein the DNN model includes a plurality of processing layers. The method includes generating a network graph of the DNN model. The method includes creating a colored network graph of the DNN model based on the identified execution order of the plurality of processing layers. The colored network graph indicates assignment of at least one memory buffer for storing at least one output of at least one processing layer. The method includes determining at least one buffer reuse overlap possibility across the plurality of processing layers. Based on the determined at least one buffer reuse overlap possibility, the method includes determining and assigning the memory required for processing the DNN model.

Abstract: Systems, apparatuses, and methods for implementing one-sided per-kernel clipping and weight transformation for neural networks are disclosed. Various parameters of a neural network are quantized from higher-bit representations to lower-bit representations to reduce memory utilization and power consumption. To exploit the effective range of quantized representations, positively biased weights are clipped and negated before convolution. Then, the results are rescaled back after convolution. A one-sided clipping technique is used for transforming weights to exploit the quantization range effectively, with the side chosen to be clipped being the biased side. This technique uses a global strategy for clipping without requiring skilled expertise. This approach allows the system to retain as much information as possible without losing unnecessary accuracy when quantizing parameters from higher-bit representations to lower-bit representations.

Abstract: Systems, methods, and devices for pruning a convolutional neural network (CNN). A subset of layers of the CNN is chosen, and for each layer of the subset of layers, how salient each filter in the layer is to an output of the CNN is determined, a subset of the filters in the layer is determined based on the salience of each filter in the layer, and the subset of filters in the layer is pruned. In some implementations, the layers of the subset of layers of the CNN are non-contiguous. In some implementations, the subset of layers includes odd numbered layers of the CNN and excludes even numbered layers of the CNN. In some implementations, the subset of layers includes even numbered layers of the CNN and excludes odd numbered layers of the CNN.

Abstract: The present disclosure disclose method and apparatus for determining memory requirement for processing a DNN model on a device, a method includes receiving a DNN model for an input, wherein the DNN model includes a plurality of processing layers. The method includes generating a network graph of the DNN model. The method includes creating a colored network graph of the DNN model based on the identified execution order of the plurality of processing layers. The colored network graph indicates assignment of at least one memory buffer for storing at least one output of at least one processing layer. The method includes determining at least one buffer reuse overlap possibility across the plurality of processing layers. Based on the determined at least one buffer reuse overlap possibility, the method includes determining and assigning the memory required for processing the DNN model.