A Generative Adversarial Network (GAN) is an unsupervised generative framework to generate a sample distribution that is identical to the data distribution. Recently, mix strategy
multi-generator/discriminator GANs have been shown to outperform single pair GANs. However, the mixed model suffers from the problem of linearly growing training time. Also, imbalanced training among generators makes it difficult to parallelize. In this paper, we propose a balanced mix-generator GAN that works in parallel by mixing multiple disjoint
generators to approximate the real distribution. The weights of the discriminator and the classifier are controlled by a balance strategy. We also present an efficient loss function, to force each generator to embrace few modes with a high probability. Our model is naturally adaptive to large parallel computation frameworks. Each generator can be trained on multiple GPUs asynchronously. We have performed extensive experiments on synthetic datasets, MNIST1000, CIFAR-10, and ImageNet. The results establish that our model can achieve the state-of-the-art performance (in terms of the modes coverage and the inception score), with significantly reduced training time. We also show that the missing mode problem can be relieved with a growing number of generators.
more »
« less
SAS: self-augmented strategy for language model pre-training.
The core of self-supervised learning for pre-training language models includes pre-training task design as well as appropriate data augmentation. Most data augmentations in language model pre-training are context-independent. A seminal contextualized augmentation was recently proposed in ELECTRA and achieved state-of-the-art performance by introducing an auxiliary generation network (generator) to produce contextualized data augmentation for the training of a main discrimination network (discriminator). This design, however, introduces extra computation cost of the genera- tor and a need to adjust the relative capability between the generator and the discriminator. In this paper, we propose a self-augmentation strategy (SAS) where a single network is utilized for both regular pre-training and contextualized data augmentation for the training in later epochs. Essentially, this strategy eliminates a separate generator and uses the single network to jointly conduct two pre-training tasks with MLM (Masked Language Modeling) and RTD (Replaced Token Detection) heads. It avoids the challenge to search for an appropriate size of the generator, which is critical to the performance as evidenced in ELECTRA and its subsequent variant models. In addition, SAS is a general strategy that can be seamlessly combined with many new techniques emerging recently or in the future, such as the disentangled attention mechanism from DeBERTa. Our experiments show that SAS outperforms ELECTRA and other state-of-the-art models in the GLUE tasks with similar or less computation cost.
more »
« less
- Award ID(s):
- 2015577
- NSF-PAR ID:
- 10351302
- Date Published:
- Journal Name:
- The Thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI) 2022
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Jovanovic, Jelena ; Chounta, Irene-Angelica ; Uhomoibhi, James ; McLaren, Bruce (Ed.)Computer-supported education studies can perform two important roles. They can allow researchers to gather important data about student learning processes, and they can help students learn more efficiently and effectively by providing automatic immediate feedback on what the students have done so far. The evaluation of student work required for both of these roles can be relatively easy in domains like math, where there are clear right answers. When text is involved, however, automated evaluations become more difficult. Natural Language Processing (NLP) can provide quick evaluations of student texts. However, traditional neural network approaches require a large amount of data to train models with enough accuracy to be useful in analyzing student responses. Typically, educational studies collect data but often only in small amounts and with a narrow focus on a particular topic. BERT-based neural network models have revolutionized NLP because they are pre-trained on very large corpora, developing a robust, contextualized understanding of the language. Then they can be “fine-tuned” on a much smaller set of data for a particular task. However, these models still need a certain base level of training data to be reasonably accurate, and that base level can exceed that provided by educational applications, which might contain only a few dozen examples. In other areas of artificial intelligence, such as computer vision, model performance on small data sets has been improved by “data augmentation” — adding scaled and rotated versions of the original images to the training set. This has been attempted on textual data; however, augmenting text is much more difficult than simply scaling or rotating images. The newly generated sentences may not be semantically similar to the original sentence, resulting in an improperly trained model. In this paper, we examine a self-augmentation method that is straightforward and shows great improvements in performance with different BERT-based models in two different languages and on two different tasks that have small data sets. We also identify the limitations of the self-augmentation procedure.more » « less
-
Full-chip thermal map estimation for multi-core commercial CPUs with generative adversarial learningIn this paper, we propose a novel transient full-chip thermal map estimation method for multi-core commercial CPU based on the data-driven generative adversarial learning method. We treat the thermal modeling problem as an image-generation problem using the generative neural networks. In stead of using traditional functional unit powers as input, the new models are directly based on the measurable real-time high level chip utilizations and thermal sensor information of commercial chips without any assumption of additional physical sensors requirement. The resulting thermal map estimation method, called {\it ThermGAN} can provide tool-accurate full-chip {\it transient} thermal maps from the given performance monitor traces of commercial off-the-shelf multi-core processors. In our work, both generator and discriminator are composed of simple convolutional layers with Wasserstein distance as loss function. ThermGAN can provide the transient and real-time thermal map without using any historical data for training and inferences, which is contrast with a recent RNN-based thermal map estimation method in which historical data is needed. Experimental results show the trained model is very accurate in thermal estimation with an average RMSE of 0.47C, namely, 0.63\% of the full-scale error. Our data further show that the speed of the model is faster than 7.5ms per inference, which is two orders of magnitude faster than the traditional finite element based thermal analysis. Furthermore, the new method is about 4x more accurate than recently proposed LSTM-based thermal map estimation method and has faster inference speed. It also achieves about 2x accuracy with much less computational cost than a state-of-the-art pre-silicon based estimation method.more » « less
-
Many applications of machine learning require a model to make accurate predictions on test examples that are distributionally different from training ones, while task-specific labels are scarce during training. An effective approach to this challenge is to pre-train a model on related tasks where data is abundant, and then fine-tune it on a downstream task of interest. While pre-training has been effective in many language and vision domains, it remains an open question how to effectively use pre-training on graph datasets. In this paper, we develop a new strategy and self-supervised methods for pre-training Graph Neural Networks (GNNs). The key to the success of our strategy is to pre-train an expressive GNN at the level of individual nodes as well as entire graphs so that the GNN can learn useful local and global representations simultaneously. We systematically study pre-training on multiple graph classification datasets. We find that naïve strategies, which pre-train GNNs at the level of either entire graphs or individual nodes, give limited improvement and can even lead to negative transfer on many downstream tasks. In contrast, our strategy avoids negative transfer and improves generalization significantly across downstream tasks, leading up to 9.4% absolute improvements in ROC-AUC over non-pre-trained models and achieving state-of-the-art performance for molecular property prediction and protein function prediction.more » « less
-
The record-breaking performance of deep neural networks (DNNs) comes with heavy parameter budgets, which leads to external dynamic random access memory (DRAM) for storage. The prohibitive energy of DRAM accesses makes it nontrivial for DNN deployment on resource-constrained devices, calling for minimizing the movements of weights and data in order to improve the energy efficiency. Driven by this critical bottleneck, we present SmartDeal, a hardware-friendly algorithm framework to trade higher-cost memory storage/access for lower-cost computation, in order to aggressively boost the storage and energy efficiency, for both DNN inference and training. The core technique of SmartDeal is a novel DNN weight matrix decomposition framework with respective structural constraints on each matrix factor, carefully crafted to unleash the hardware-aware efficiency potential. Specifically, we decompose each weight tensor as the product of a small basis matrix and a large structurally sparse coefficient matrix whose nonzero elements are readily quantized to the power-of-2. The resulting sparse and readily quantized DNNs enjoy greatly reduced energy consumption in data movement as well as weight storage, while incurring minimal overhead to recover the original weights thanks to the required sparse bit-operations and cost-favorable computations. Beyond inference, we take another leap to embrace energy-efficient training, by introducing several customized techniques to address the unique roadblocks arising in training while preserving the SmartDeal structures. We also design a dedicated hardware accelerator to fully utilize the new weight structure to improve the real energy efficiency and latency performance. We conduct experiments on both vision and language tasks, with nine models, four datasets, and three settings (inference-only, adaptation, and fine-tuning). Our extensive results show that 1) being applied to inference, SmartDeal achieves up to 2.44x improvement in energy efficiency as evaluated using real hardware implementations and 2) being applied to training, SmartDeal can lead to 10.56x and 4.48x reduction in the storage and the training energy cost, respectively, with usually negligible accuracy loss, compared to state-of-the-art training baselines. Our source codes are available at: https://github.com/VITA-Group/SmartDeal.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1609/aaai.v36i10.21412
