We introduce QUANTISENC, a fully-configurable digital spiking neuromorphic hardware to optimize performance and power consumption of spiking neural networks (SNNs). QUANTISENC introduces two key contributions. First, ...
详细信息
ISBN:
(纸本)9798350387186;9798350387179
We introduce QUANTISENC, a fully-configurable digital spiking neuromorphic hardware to optimize performance and power consumption of spiking neural networks (SNNs). QUANTISENC introduces two key contributions. First, it allows the user to set separate quantization and precision policies for the synaptic weights and the internal state variables of neurons to optimize the design based on the precision needed for a target SNN model and the dataset used for training. This reduces the quantization error. Second, in addition to using static design parameters, QUANTISENC also allows to dynamically configure neuron parameters via programming its configuration registers. This allows the user to fine-tune performance and power consumption even after a design is implemented on silicon. Using open-source datasets, we show improvement in area, power, and performance over several state-of-the-art designs.
For image compression applications where the information sink is not a person but a computer algorithm, the image encoder should control the encoding process in such a way that the important and relevant features of t...
详细信息
ISBN:
(纸本)9781457713033
For image compression applications where the information sink is not a person but a computer algorithm, the image encoder should control the encoding process in such a way that the important and relevant features of the image are preserved after compression. In this paper, our goal is to preserve the strongest SIFT features for JPEG-encoded images. We analyze the relevant characteristics of SIFT features and categorize the image Macroblocks into several groups. Then we propose a novel rate-distortion model which is based on the SIFT feature matching score. The dependency between the quantization table in the JPEG file and the common Lagrange multiplier is obtained from a training image database. Then for a given image quality we exploit this relationship to perform R-D optimization for each group. Our results show that the proposed algorithm achieves better feature preservation when compared to standard JPEG encoding. The proposed approach is fully standard compatible.
In this paper, we propose a low complexity decoder architecture for low-density parity-check (LDPC) codes using a variable quantization scheme as well as an efficient highly-parallel decoding scheme. In the sum-produc...
详细信息
In this paper, we propose a low complexity decoder architecture for low-density parity-check (LDPC) codes using a variable quantization scheme as well as an efficient highly-parallel decoding scheme. In the sum-product algorithm for decoding LDPC codes, the finite precision implementations have an important tradeoff between decoding performance and hardware complexity caused by two dominant area-consuming factors: one is the memory for updated messages storage and the other is the look-up table (LUT) for implementation of the nonlinear function Psi(x). The proposed variable quantization schemes offer a large reduction in the hardware complexities for LUT and memory. Also, an efficient highly- parallel decoder architecture for quasi-cyclic (QC) LDPC codes can be implemented with the reduced hardware complexity by using the partially block overlapped decoding scheme and the minimized power consumption by reducing the total number of memory accesses for updated messages. For (3, 6) QC LDPC codes, our proposed schemes in implementing the highly- parallel decoder architecture offer a great reduction of implementation area by 33% for memory area and approximately by 28% for the check node unit and variable node unit computation units without significant performance degradation. Also, the memory accesses are reduced by 20%.
In this paper, we present a novel technique for the suppression of the return channel in a multisensor distributed Wyner-Ziv video coding system. The proposed technique relies on a cross-layer approach that takes into...
详细信息
ISBN:
(纸本)9781424417513
In this paper, we present a novel technique for the suppression of the return channel in a multisensor distributed Wyner-Ziv video coding system. The proposed technique relies on a cross-layer approach that takes into account the transmission channel conditions on one hand, and the content of the captured video scenes on the other, in order to allocate different transmission rates and dynamically vary the quantization parameter for each user. Simulation results show significant improvement in the average system performance compared to a traditional system where all users are assigned equal channel resources and a fixed number of quantization levels.
In this paper, we present a novel technique for the suppression of the return channel in a multisensor distributed Wyner-Ziv video coding system. The proposed technique relies on a cross-layer approach that takes into...
详细信息
In this paper, we present a novel technique for the suppression of the return channel in a multisensor distributed Wyner-Ziv video coding system. The proposed technique relies on a cross-layer approach that takes into account the transmission channel conditions on one hand, and the content of the captured video scenes on the other, in order to allocate different transmission rates and dynamically vary the quantization parameter for each user. Simulation results show significant improvement in the average system performance compared to a traditional system where all users are assigned equal channel resources and a fixed number of quantization levels.
暂无评论