Two challenges make it difficult to cluster on-line AECG (Ambulatory Electrocardiogram) data. they are huge amount ECG (Electrocardiogram) waveforms and high dimension vector to describe individual ECG waveforms. A ne...
详细信息
Convolutional neural networks (CNN) applications range from recognition and reasoning (such as handwriting recognition, facial expression recognition and video surveillance) to intelligent text applications such as se...
详细信息
ISBN:
(纸本)9781450300520
Convolutional neural networks (CNN) applications range from recognition and reasoning (such as handwriting recognition, facial expression recognition and video surveillance) to intelligent text applications such as semantic text analysis and natural language processing applications. Two key observations drive the design of a new architecture for CNN. First, CNN workloads exhibit a widely varying mix of three types of parallelism: parallelism within a convolution operation, intra-output parallelism where multiple input sources (features) are combined to create a single output, and inter-output parallelism where multiple, independent outputs (features) are computed simultaneously. Workloads differ significantly across different CNN applications, and across different layers of a CNN. Second, the number of processing elements in an architecture continues to scale (as per Moore's law) much faster than off-chip memory bandwidth (or pin-count) of chips. Based on these two observations, we show that for a given number of processing elements and off-chip memory bandwidth, a new CNN hardware architecture that dynamically configures the hardware on-the-fly to match the specific mix of parallelism in a given workload gives the best throughput performance. Our CNN compiler automatically translates high abstraction network specification into a parallel microprogram (a sequence of low-level VLIW instructions) that is mapped, scheduled and executed by the coprocessor. Compared to a 2.3 GHz quad-core, dual socket Intel Xeon, 1.35 GHz C870 GPU, and a 200 MHz FPGA implementation, our 120 MHz dynamically configurable architecture is 4x to 8x faster. this is the first CNN architecture to achieve real-time video stream processing (25 to 30 frames per second) on a wide range of object detection and recognition tasks.
In view of serious cache missing of each macro block of MPEG-4 (Moving Picture Experts Group-4) video encoder in transmission and the video sequence frame rate is low, We proposed the motion estimation algorithm optim...
详细信息
Inspired information processing in biology immune system is a highly parallel and distributed intelligent computation which has learning, memory, and associative retrieval capabilities. In this paper, we design a netw...
详细信息
the analysis and characterization of biomedical image data is a complex procedure involving several processing phases, like data acquisition, preprocessing, segmentation, feature extraction and classification. the pro...
详细信息
Because small mobile devices do not include a floating-point processing unit due to their circuit complexity and power consumption, there exist some disconnections or noises caused by lengthening the decoding time of ...
详细信息
ISBN:
(纸本)9788988678299
Because small mobile devices do not include a floating-point processing unit due to their circuit complexity and power consumption, there exist some disconnections or noises caused by lengthening the decoding time of audio data. this study proposes an algorithm that improves the decoding rate through the conversion of a real operation process to an integer operation process applied in an MPEG-4 audio decoding process in order to solve this problem. Moreover, this study implements the software that plays the MPEG-4 AAC data in real-time in embedded Linux based mobile devices and verifies the possibility of the replay of MPEG-4 audio data in its error range through the results of related experiments.
Key management in sensor networks is a challenging issue due to the limited resource constraints of sensors. Many proposed schemes are based on key pre-distribution using pre-deployment knowledge that is difficult to ...
详细信息
the Multi-agents computing paradigm offers support for large scale, widely distributed, high-performance computational systems. Several of such architectures and frameworks have been developed aimed at primarily compu...
详细信息
暂无评论