اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدLUTNet: Learning FPGA Configurations for Highly Efficient Neural Network Inference
106
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Research has shown that deep neural networks contain significant redundancy, and thus that high classification accuracy can be achieved even when weights and activations are quantized down to binary values. Network binarization on FPGAs greatly increases area efficiency by replacing resource-hungry multipliers with lightweight XNOR gates. However, an FPGAs fundamental building block, the K-LUT, is capable of implementing far more than an XNOR: it can perform any K-input Boolean operation. Inspired by this observation, we propose LUTNet, an end-to-end hardware-software framework for the construction of area-efficient FPGA-based neural network accelerators using the native LUTs as inference operators. We describe the realization of both unrolled and tiled LUTNet architectures, with the latter facilitating smaller, less power-hungry deployment over the former while sacrificing area and energy efficiency along with throughput. For both varieties, we demonstrate that the exploitation of LUT flexibility allows for far heavier pruning than possible in prior works, resulting in significant area savings while achieving comparable accuracy. Against the state-of-the-art binarized neural network implementation, we achieve up to twice the area efficiency for several standard network models when inferencing popular datasets. We also demonstrate that even greater energy efficiency improvements are obtainable.
قيم البحث
اقرأ أيضاً
Research has shown that deep neural networks contain significant redundancy, and that high classification accuracies can be achieved even when weights and activations are quantised down to binary values. Network binarisation on FPGAs greatly increase
s area efficiency by replacing resource-hungry multipliers with lightweight XNOR gates. However, an FPGAs fundamental building block, the K-LUT, is capable of implementing far more than an XNOR: it can perform any K-input Boolean operation. Inspired by this observation, we propose LUTNet, an end-to-end hardware-software framework for the construction of area-efficient FPGA-based neural network accelerators using the native LUTs as inference operators. We demonstrate that the exploitation of LUT flexibility allows for far heavier pruning than possible in prior works, resulting in significant area savings while achieving comparable accuracy. Against the state-of-the-art binarised neural network implementation, we achieve twice the area efficiency for several standard network models when inferencing popular datasets. We also demonstrate that even greater energy efficiency improvements are obtainable.
Automated neural network design has received ever-increasing attention with the evolution of deep convolutional neural networks (CNNs), especially involving their deployment on embedded and mobile platforms. One of the biggest problems that neural ar
chitecture search (NAS) confronts is that a large number of candidate neural architectures are required to train, using, for instance, reinforcement learning and evolutionary optimisation algorithms, at a vast computation cost. Even recent differentiable neural architecture search (DNAS) samples a small number of candidate neural architectures based on the probability distribution of learned architecture parameters to select the final neural architecture. To address this computational complexity issue, we introduce a novel emph{architecture parameterisation} based on scaled sigmoid function, and propose a general emph{Differentiable Neural Architecture Learning} (DNAL) method to optimize the neural architecture without the need to evaluate candidate neural networks. Specifically, for stochastic supernets as well as conventional CNNs, we build a new channel-wise module layer with the architecture components controlled by a scaled sigmoid function. We train these neural network models from scratch. The network optimization is decoupled into the weight optimization and the architecture optimization. We address the non-convex optimization problem of neural architecture by the continuous scaled sigmoid method with convergence guarantees. Extensive experiments demonstrate our DNAL method delivers superior performance in terms of neural architecture search cost. The optimal networks learned by DNAL surpass those produced by the state-of-the-art methods on the benchmark CIFAR-10 and ImageNet-1K dataset in accuracy, model size and computational complexity.
We present PrecisionBatching, a quantized inference algorithm for speeding up neural network execution on traditional hardware platforms at low bitwidths without the need for retraining or recalibration. PrecisionBatching decomposes a neural network
into individual bitlayers and accumulates them using fast 1-bit operations while maintaining activations in full precision. PrecisionBatching not only facilitates quantized inference at low bitwidths (< 8 bits) without the need for retraining/recalibration, but also 1) enables traditional hardware platforms the ability to realize inference speedups at a finer granularity of quantization (e.g: 1-16 bit execution) and 2) allows accuracy and speedup tradeoffs at runtime by exposing the number of bitlayers to accumulate as a tunable parameter. Across a variety of applications (MNIST, language modeling, natural language inference) and neural network architectures (fully connected, RNN, LSTM), PrecisionBatching yields end-to-end speedups of over 8x on a GPU within a < 1% error margin of the full precision baseline, outperforming traditional 8-bit quantized inference by over 1.5x-2x at the same error tolerance.
Sparse neural networks can greatly facilitate the deployment of neural networks on resource-constrained platforms as they offer compact model sizes while retaining inference accuracy. Because of the sparsity in parameter matrices, sparse neural netwo
rks can, in principle, be exploited in accelerator architectures for improved energy-efficiency and latency. However, to realize these improvements in practice, there is a need to explore sparsity-aware hardware-software co-design. In this paper, we propose a novel silicon photonics-based sparse neural network inference accelerator called SONIC. Our experimental analysis shows that SONIC can achieve up to 5.8x better performance-per-watt and 8.4x lower energy-per-bit than state-of-the-art sparse electronic neural network accelerators; and up to 13.8x better performance-per-watt and 27.6x lower energy-per-bit than the best known photonic neural network accelerators.
Efficient processing of large-scale time series data is an intricate problem in machine learning. Conventional sensor signal processing pipelines with hand engineered feature extraction often involve huge computational cost with high dimensional data
. Deep recurrent neural networks have shown promise in automated feature learning for improved time-series processing. However, generic deep recurrent models grow in scale and depth with increased complexity of the data. This is particularly challenging in presence of high dimensional data with temporal and spatial characteristics. Consequently, this work proposes a novel deep cellular recurrent neural network (DCRNN) architecture to efficiently process complex multi-dimensional time series data with spatial information. The cellular recurrent architecture in the proposed model allows for location-aware synchronous processing of time series data from spatially distributed sensor signal sources. Extensive trainable parameter sharing due to cellularity in the proposed architecture ensures efficiency in the use of recurrent processing units with high-dimensional inputs. This study also investigates the versatility of the proposed DCRNN model for classification of multi-class time series data from different application domains. Consequently, the proposed DCRNN architecture is evaluated using two time-series datasets: a multichannel scalp EEG dataset for seizure detection, and a machine fault detection dataset obtained in-house. The results suggest that the proposed architecture achieves state-of-the-art performance while utilizing substantially less trainable parameters when compared to comparable methods in the literature.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
