Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userAdaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers
307
0
0.0
(
0
)
Added by
Xishan Zhang
Publication date
2019
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers. Recent emerged quantization technique has been applied to inference of deep neural networks for fast and efficient execution. However, directly applying quantization in training can cause significant accuracy loss, thus remaining an open challenge.
rate research
Read More
Todays deep learning models are primarily trained on CPUs and GPUs. Although these models tend to have low error, they consume high power and utilize large amount of memory owing to double precision floating point learning parameters. Beyond the Moores law, a significant portion of deep learning tasks would run on edge computing systems, which will form an indispensable part of the entire computation fabric. Subsequently, training deep learning models for such systems will have to be tailored and adopted to generate models that have the following desirable characteristics: low error, low memory, and low power. We believe that deep neural networks (DNNs), where learning parameters are constrained to have a set of finite discrete values, running on neuromorphic computing systems would be instrumental for intelligent edge computing systems having these desirable characteristics. To this extent, we propose the Combinatorial Neural Network Training Algorithm (CoNNTrA), that leverages a coordinate gradient descent-based approach for training deep learning models with finite discrete learning parameters. Next, we elaborate on the theoretical underpinnings and evaluate the computational complexity of CoNNTrA. As a proof of concept, we use CoNNTrA to train deep learning models with ternary learning parameters on the MNIST, Iris and ImageNet data sets and compare their performance to the same models trained using Backpropagation. We use following performance metrics for the comparison: (i) Training error; (ii) Validation error; (iii) Memory usage; and (iv) Training time. Our results indicate that CoNNTrA models use 32x less memory and have errors at par with the Backpropagation models.
Graph neural networks (GNNs) are processing architectures that exploit graph structural information to model representations from network data. Despite their success, GNNs suffer from sub-optimal generalization performance given limited training data, referred to as over-fitting. This paper proposes Topology Adaptive Edge Dropping (TADropEdge) method as an adaptive data augmentation technique to improve generalization performance and learn robust GNN models. We start by explicitly analyzing how random edge dropping increases the data diversity during training, while indicating i.i.d. edge dropping does not account for graph structural information and could result in noisy augmented data degrading performance. To overcome this issue, we consider graph connectivity as the key property that captures graph topology. TADropEdge incorporates this factor into random edge dropping such that the edge-dropped subgraphs maintain similar topology as the underlying graph, yielding more satisfactory data augmentation. In particular, TADropEdge first leverages the graph spectrum to assign proper weights to graph edges, which represent their criticality for establishing the graph connectivity. It then normalizes the edge weights and drops graph edges adaptively based on their normalized weights. Besides improving generalization performance, TADropEdge reduces variance for efficient training and can be applied as a generic method modular to different GNN models. Intensive experiments on real-life and synthetic datasets corroborate theory and verify the effectiveness of the proposed method.
In a previous work we have detailed the requirements to obtain a maximal performance benefit by implementing fully connected deep neural networks (DNN) in form of arrays of resistive devices for deep learning. This concept of Resistive Processing Unit (RPU) devices we extend here towards convolutional neural networks (CNNs). We show how to map the convolutional layers to RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed due to analog nature of the computations performed on the arrays effect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of RPU approach for large class of neural network architectures.
Convolutional neural networks have achieved astonishing results in different application areas. Various methods that allow us to use these models on mobile and embedded devices have been proposed. Especially binary neural networks are a promising approach for devices with low computational power. However, training accurate binary models from scratch remains a challenge. Previous work often uses prior knowledge from full-precision models and complex training strategies. In our work, we focus on increasing the performance of binary neural networks without such prior knowledge and a much simpler training strategy. In our experiments we show that we are able to achieve state-of-the-art results on standard benchmark datasets. Further, to the best of our knowledge, we are the first to successfully adopt a network architecture with dense connections for binary networks, which lets us improve the state-of-the-art even further.
Reduced precision computation for deep neural networks is one of the key areas addressing the widening compute gap driven by an exponential growth in model size. In recent years, deep learning training has largely migrated to 16-bit precision, with significant gains in performance and energy efficiency. However, attempts to train DNNs at 8-bit precision have met with significant challenges because of the higher precision and dynamic range requirements of back-propagation. In this paper, we propose a method to train deep neural networks using 8-bit floating point representation for weights, activations, errors, and gradients. In addition to reducing compute precision, we also reduced the precision requirements for the master copy of weights from 32-bit to 16-bit. We demonstrate state-of-the-art accuracy across multiple data sets (imagenet-1K, WMT16) and a broader set of workloads (Resnet-18/34/50, GNMT, Transformer) than previously reported. We propose an enhanced loss scaling method to augment the reduced subnormal range of 8-bit floating point for improved error propagation. We also examine the impact of quantization noise on generalization and propose a stochastic rounding technique to address gradient noise. As a result of applying all these techniques, we report slightly higher validation accuracy compared to full precision baseline.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
