Chen, Enhong

Image Denoising and Inpainting with Deep Neural Networks

Neural Information Processing Systems

We present a novel approach to low-level vision problems that combines sparse coding and deep networks pre-trained with denoising auto-encoder (DA). We propose an alternative training scheme that successfully adapts DA, originally designed for unsupervised feature learning, to the tasks of image denoising and blind inpainting. Our method achieves state-of-the-art performance in the image denoising task. More importantly, in blind image inpainting task, the proposed method provides solutions to some complex problems that have not been tackled before. Specifically, we can automatically remove complex patterns like superimposed text from an image, rather than simple patterns like pixels missing at random.

Neural Architecture Optimization

Neural Information Processing Systems

Automatic neural architecture design has shown its potential in discovering powerful neural network architectures. Existing methods, no matter based on reinforcement learning or evolutionary algorithms (EA), conduct architecture search in a discrete space, which is highly inefficient. In this paper, we propose a simple and efficient method to automatic neural architecture design based on continuous optimization. We call this new approach neural architecture optimization (NAO). There are three key components in our proposed approach: (1) An encoder embeds/maps neural network architectures into a continuous space.

Domain Adaption for Knowledge Tracing Artificial Intelligence

With the rapid development of online education system, knowledge tracing which aims at predicting students' knowledge state is becoming a critical and fundamental task in personalized education. Traditionally, existing methods are domain-specified. However, there are a larger number of domains (e.g., subjects, schools) in the real world and the lacking of data in some domains, how to utilize the knowledge and information in other domains to help train a knowledge tracing model for target domains is increasingly important. We refer to this problem as domain adaptation for knowledge tracing (DAKT) which contains two aspects: (1) how to achieve great knowledge tracing performance in each domain. (2) how to transfer good performed knowledge tracing model between domains. To this end, in this paper, we propose a novel adaptable framework, namely adaptable knowledge tracing (AKT) to address the DAKT problem. Specifically, for the first aspect, we incorporate the educational characteristics (e.g., slip, guess, question texts) based on the deep knowledge tracing (DKT) to obtain a good performed knowledge tracing model. For the second aspect, we propose and adopt three domain adaptation processes. First, we pre-train an auto-encoder to select useful source instances for target model training. Second, we minimize the domain-specific knowledge state distribution discrepancy under maximum mean discrepancy (MMD) measurement to achieve domain adaptation. Third, we adopt fine-tuning to deal with the problem that the output dimension of source and target domain are different to make the model suitable for target domains. Extensive experimental results on two private datasets and seven public datasets clearly prove the effectiveness of AKT for great knowledge tracing performance and its superior transferable ability.

Efficient Pure Exploration in Adaptive Round model

Neural Information Processing Systems

In the adaptive setting, many multi-armed bandit applications allow the learner to adaptively draw samples and adjust sampling strategy in rounds. In many real applications, not only the query complexity but also the round complexity need to be optimized. In this paper, we study both PAC and exact top-$k$ arm identification problems and design efficient algorithms considering both round complexity and query complexity. For PAC problem, we achieve optimal query complexity and use only $O(\log_{\frac{k}{\delta}} *(n))$ rounds, which matches the lower bound of round complexity, while most of existing works need $\Theta(\log \frac{n}{k})$ rounds. For exact top-$k$ arm identification, we improve the round complexity factor from $\log n$ to $\log_{\frac{1}{\delta}} *(n)$, and achieve near optimal query complexity.

Deep Technology Tracing for High-tech Companies Machine Learning

Technological change and innovation are vitally important, especially for high-tech companies. However, factors influencing their future research and development (R&D) trends are both complicated and various, leading it a quite difficult task to make technology tracing for high-tech companies. To this end, in this paper, we develop a novel data-driven solution, i.e., Deep Technology Forecasting (DTF) framework, to automatically find the most possible technology directions customized to each high-tech company. Specially, DTF consists of three components: Potential Competitor Recognition (PCR), Collaborative Technology Recognition (CTR), and Deep Technology Tracing (DTT) neural network. For one thing, PCR and CTR aim to capture competitive relations among enterprises and collaborative relations among technologies, respectively. For another, DTT is designed for modeling dynamic interactions between companies and technologies with the above relations involved. Finally, we evaluate our DTF framework on real-world patent data, and the experimental results clearly prove that DTF can precisely help to prospect future technology emphasis of companies by exploiting hybrid factors.

Variance Reduced Local SGD with Lower Communication Complexity Machine Learning

To accelerate the training of machine learning models, distributed stochastic gradient descent (SGD) and its variants have been widely adopted, which apply multiple workers in parallel to speed up training. Among them, Local SGD has gained much attention due to its lower communication cost. Nevertheless, when the data distribution on workers is non-identical, Local SGD requires $O(T^{\frac{3}{4}} N^{\frac{3}{4}})$ communications to maintain its \emph{linear iteration speedup} property, where $T$ is the total number of iterations and $N$ is the number of workers. In this paper, we propose Variance Reduced Local SGD (VRL-SGD) to further reduce the communication complexity. Benefiting from eliminating the dependency on the gradient variance among workers, we theoretically prove that VRL-SGD achieves a \emph{linear iteration speedup} with a lower communication complexity $O(T^{\frac{1}{2}} N^{\frac{3}{2}})$ even if workers access non-identical datasets. We conduct experiments on three machine learning tasks, and the experimental results demonstrate that VRL-SGD performs impressively better than Local SGD when the data among workers are quite diverse.

Estimating Early Fundraising Performance of Innovations via Graph-based Market Environment Model Machine Learning

Well begun is half done. In the crowdfunding market, the early fundraising performance of the project is a concerned issue for both creators and platforms. However, estimating the early fundraising performance before the project published is very challenging and still under-explored. To that end, in this paper, we present a focused study on this important problem in a market modeling view. Specifically, we propose a Graph-based Market Environment model (GME) for estimating the early fundraising performance of the target project by exploiting the market environment. In addition, we discriminatively model the market competition and market evolution by designing two graph-based neural network architectures and incorporating them into the joint optimization stage. Finally, we conduct extensive experiments on the real-world crowdfunding data collected from The experimental results clearly demonstrate the effectiveness of our proposed model for modeling and estimating the early fundraising performance of the target project.

Fine-Tuning by Curriculum Learning for Non-Autoregressive Neural Machine Translation Machine Learning

Non-autoregressive translation (NAT) models remove the dependence on previous target tokens and generate all target tokens in parallel, resulting in significant inference speedup but at the cost of inferior translation accuracy compared to autoregressive translation (AT) models. Considering that AT models have higher accuracy and are easier to train than NAT models, and both of them share the same model configurations, a natural idea to improve the accuracy of NAT models is to transfer a well-trained AT model to an NAT model through fine-tuning. However, since AT and NAT models differ greatly in training strategy, straightforward fine-tuning does not work well. In this work, we introduce curriculum learning into fine-tuning for NAT. Specifically, we design a curriculum in the fine-tuning process to progressively switch the training from autoregressive generation to non-autoregressive generation. Experiments on four benchmark translation datasets show that the proposed method achieves good improvement (more than $1$ BLEU score) over previous NAT baselines in terms of translation accuracy, and greatly speed up (more than $10$ times) the inference process over AT baselines.

Long-term Joint Scheduling for Urban Traffic Artificial Intelligence

Recently, the traffic congestion in modern cities has become a growing worry for the residents. As presented in Baidu traffic report, the commuting stress index has reached surprising 1.973 in Beijing during rush hours, which results in longer trip time and increased vehicular queueing. Previous works have demonstrated that by reasonable scheduling, e.g, rebalancing bike-sharing systems and optimized bus transportation, the traffic efficiency could be significantly improved with little resource consumption. However, there are still two disadvantages that restrict their performance: (1) they only consider single scheduling in a short time, but ignoring the layout after first reposition, and (2) they only focus on the single transport. However, the multi-modal characteristics of urban public transportation are largely under-exploited. In this paper, we propose an efficient and economical multi-modal traffic scheduling scheme named JLRLS based on spatio -temporal prediction, which adopts reinforcement learning to obtain optimal long-term and joint schedule. In JLRLS, we combines multiple transportation to conduct scheduling by their own characteristics, which potentially helps the system to reach the optimal performance. Our implementation of an example by PaddlePaddle is available at, with an explaining video at

Understanding and Improving One-shot Neural Architecture Optimization Machine Learning

The ability of accurately ranking candidate architectures is the key to the performance of neural architecture search~(NAS). One-shot NAS is proposed to cut the expense but shows inferior performance against conventional NAS and is not adequately stable. We find that the ranking correlation between architectures under one-shot training and the ones under stand-alone training is poor, which misleads the algorithm to discover better architectures. We conjecture that this is owing to the gaps between one-shot training and stand-alone complete training. In this work, we empirically investigate several main factors that lead to the gaps and so weak ranking correlation. We then propose NAO-V2 to alleviate such gaps where we: (1) Increase the average updates for individual architecture to a relatively adequate extent. (2) Encourage more updates for large and complex architectures than small and simple architectures to balance them by sampling architectures in proportion to their model sizes. (3) Make the one-shot training of the supernet independent at each iteration. Comprehensive experiments verify that our proposed method is effective and robust. It leads to a more stable search that all the top architectures perform well enough compared to baseline methods. The final discovered architecture shows significant improvements against baselines with a test error rate of 2.60% on CIFAR-10 and top-1 accuracy of 74.4% on ImageNet under the mobile setting. Code and model checkpoints are publicly available at