arXiv subjects

Chao Deng

Publications and source records attributed to Chao Deng.

At least 91 records · Page 5Linked to original sources

arXiv2023

Federated Learning over Coupled Graphs

Graphs are widely used to represent the relations among entities. When one owns the complete data, an entire graph can be easily built, therefore performing analysis on the graph is straightforward. However, in many scenarios, it is impractical to centralize the data due to data privacy concerns. An organization or party only keeps a part of the whole graph data, i.e., graph data is isolated from different parties. Recently, Federated Learning (FL) has been proposed to solve the data isolation issue, mainly for Euclidean data. It is still a challenge to apply FL on graph data because graphs contain topological information which is notorious for its non-IID nature and is hard to partition. In this work, we propose a novel FL framework for graph data, FedCog, to efficiently handle coupled graphs that are a kind of distributed graph data, but widely exist in a variety of real-world applications such as mobile carriers' communication networks and banks' transaction networks. We theoretically prove the correctness and security of FedCog. Experimental results demonstrate that our method FedCog significantly outperforms traditional FL methods on graphs. Remarkably, our FedCog improves the accuracy of node classification tasks by up to 14.7%.

cs.LG↗

arXiv2022

Learning to Weight Samples for Dynamic Early-exiting Networks

Early exiting is an effective paradigm for improving the inference efficiency of deep networks. By constructing classifiers with varying resource demands (the exits), such networks allow easy samples to be output at early exits, removing the need for executing deeper layers. While existing works mainly focus on the architectural design of multi-exit networks, the training strategies for such models are largely left unexplored. The current state-of-the-art models treat all samples the same during training. However, the early-exiting behavior during testing has been ignored, leading to a gap between training and testing. In this paper, we propose to bridge this gap by sample weighting. Intuitively, easy samples, which generally exit early in the network during inference, should contribute more to training early classifiers. The training of hard samples (mostly exit from deeper layers), however, should be emphasized by the late classifiers. Our work proposes to adopt a weight prediction network to weight the loss of different training samples at each exit. This weight prediction network and the backbone model are jointly optimized under a meta-learning framework with a novel optimization objective. By bringing the adaptive behavior during inference into the training phase, we show that the proposed weighting mechanism consistently improves the trade-off between classification accuracy and inference efficiency. Code is available at https://github.com/LeapLabTHU/L2W-DEN.

cs.CV↗

arXiv2022

Meta Auxiliary Learning for Low-resource Spoken Language Understanding

Spoken language understanding (SLU) treats automatic speech recognition (ASR) and natural language understanding (NLU) as a unified task and usually suffers from data scarcity. We exploit an ASR and NLU joint training method based on meta auxiliary learning to improve the performance of low-resource SLU task by only taking advantage of abundant manual transcriptions of speech data. One obvious advantage of such method is that it provides a flexible framework to implement a low-resource SLU training task without requiring access to any further semantic annotations. In particular, a NLU model is taken as label generation network to predict intent and slot tags from texts; a multi-task network trains ASR task and SLU task synchronously from speech; and the predictions of label generation network are delivered to the multi-task network as semantic targets. The efficiency of the proposed algorithm is demonstrated with experiments on the public CATSLU dataset, which produces more suitable ASR hypotheses for the downstream NLU task.

eess.AS↗

arXiv2022

A CTC Triggered Siamese Network with Spatial-Temporal Dropout for Speech Recognition

Siamese networks have shown effective results in unsupervised visual representation learning. These models are designed to learn an invariant representation of two augmentations for one input by maximizing their similarity. In this paper, we propose an effective Siamese network to improve the robustness of End-to-End automatic speech recognition (ASR). We introduce spatial-temporal dropout to support a more violent disturbance for Siamese-ASR framework. Besides, we also relax the similarity regularization to maximize the similarities of distributions on the frames that connectionist temporal classification (CTC) spikes occur rather than on all of them. The efficiency of the proposed architecture is evaluated on two benchmarks, AISHELL-1 and Librispeech, resulting in 7.13% and 6.59% relative character error rate (CER) and word error rate (WER) reductions respectively. Analysis shows that our proposed approach brings a better uniformity for the trained model and enlarges the CTC spikes obviously.

eess.AS↗

arXiv2022

A Semantic Consistency Feature Alignment Object Detection Model Based on Mixed-Class Distribution Metrics

Unsupervised domain adaptation is critical in various computer vision tasks, such as object detection, instance segmentation, etc. They attempt to reduce domain bias-induced performance degradation while also promoting model application speed. Previous works in domain adaptation object detection attempt to align image-level and instance-level shifts to eventually minimize the domain discrepancy, but they may align single-class features to mixed-class features in image-level domain adaptation because each image in the object detection task may be more than one class and object. In order to achieve single-class with single-class alignment and mixed-class with mixed-class alignment, we treat the mixed-class of the feature as a new class and propose a mixed-classes $H-divergence$ for object detection to achieve homogenous feature alignment and reduce negative transfer. Then, a Semantic Consistency Feature Alignment Model (SCFAM) based on mixed-classes $H-divergence$ was also presented. To improve single-class and mixed-class semantic information and accomplish semantic separation, the SCFAM model proposes Semantic Prediction Models (SPM) and Semantic Bridging Components (SBC). And the weight of the pix domain discriminator loss is then changed based on the SPM result to reduce sample imbalance. Extensive unsupervised domain adaption experiments on widely used datasets illustrate our proposed approach's robust object detection in domain bias settings.

cs.CV↗

arXiv2022

OPAL: Occlusion Pattern Aware Loss for Unsupervised Light Field Disparity Estimation

Light field disparity estimation is an essential task in computer vision with various applications. Although supervised learning-based methods have achieved both higher accuracy and efficiency than traditional optimization-based methods, the dependency on ground-truth disparity for training limits the overall generalization performance not to say for real-world scenarios where the ground-truth disparity is hard to capture. In this paper, we argue that unsupervised methods can achieve comparable accuracy, but, more importantly, much higher generalization capacity and efficiency than supervised methods. Specifically, we present the Occlusion Pattern Aware Loss, named OPAL, which successfully extracts and encodes the general occlusion patterns inherent in the light field for loss calculation. OPAL enables: i) accurate and robust estimation by effectively handling occlusions without using any ground-truth information for training and ii) much efficient performance by significantly reducing the network parameters required for accurate inference. Besides, a transformer-based network and a refinement module are proposed for achieving even more accurate results. Extensive experiments demonstrate our method not only significantly improves the accuracy compared with the SOTA unsupervised methods, but also possesses strong generalization capacity, even for real-world data, compared with supervised methods. Our code will be made publicly available.

cs.CV↗

arXiv2022

GenAD: General Representations of Multivariate Time Seriesfor Anomaly Detection

The reliability of wireless base stations in China Mobile is of vital importance, because the cell phone users are connected to the stations and the behaviors of the stations are directly related to user experience. Although the monitoring of the station behaviors can be realized by anomaly detection on multivariate time series, due to complex correlations and various temporal patterns of multivariate series in large-scale stations, building a general unsupervised anomaly detection model with a higher F1-score remains a challenging task. In this paper, we propose a General representation of multivariate time series for Anomaly Detection(GenAD). First, we pre-train a general model on large-scale wireless base stations with self-supervision, which can be easily transferred to a specific station anomaly detection with a small amount of training data. Second, we employ Multi-Correlation Attention and Time-Series Attention to represent the correlations and temporal patterns of the stations. With the above innovations, GenAD increases F1-score by total 9% on real-world datasets in China Mobile, while the performance does not significantly degrade on public datasets with only 10% of the training data.

cs.NI↗

arXiv2021

Adaptive Multi-receptive Field Spatial-Temporal Graph Convolutional Network for Traffic Forecasting

Mobile network traffic forecasting is one of the key functions in daily network operation. A commercial mobile network is large, heterogeneous, complex and dynamic. These intrinsic features make mobile network traffic forecasting far from being solved even with recent advanced algorithms such as graph convolutional network-based prediction approaches and various attention mechanisms, which have been proved successful in vehicle traffic forecasting. In this paper, we cast the problem as a spatial-temporal sequence prediction task. We propose a novel deep learning network architecture, Adaptive Multi-receptive Field Spatial-Temporal Graph Convolutional Networks (AMF-STGCN), to model the traffic dynamics of mobile base stations. AMF-STGCN extends GCN by (1) jointly modeling the complex spatial-temporal dependencies in mobile networks, (2) applying attention mechanisms to capture various Receptive Fields of heterogeneous base stations, and (3) introducing an extra decoder based on a fully connected deep network to conquer the error propagation challenge with multi-step forecasting. Experiments on four real-world datasets from two different domains consistently show AMF-STGCN outperforms the state-of-the-art methods.

cs.LG↗

arXiv2021

10-mega pixel snapshot compressive imaging with a hybrid coded aperture

High resolution images are widely used in our daily life, whereas high-speed video capture is challenging due to the low frame rate of cameras working at the high resolution mode. Digging deeper, the main bottleneck lies in the low throughput of existing imaging systems. Towards this end, snapshot compressive imaging (SCI) was proposed as a promising solution to improve the throughput of imaging systems by compressive sampling and computational reconstruction. During acquisition, multiple high-speed images are encoded and collapsed to a single measurement. After this, algorithms are employed to retrieve the video frames from the coded snapshot. Recently developed Plug-and-Play (PnP) algorithms make it possible for SCI reconstruction in large-scale problems. However, the lack of high-resolution encoding systems still precludes SCI's wide application. In this paper, we build a novel hybrid coded aperture snapshot compressive imaging (HCA-SCI) system by incorporating a dynamic liquid crystal on silicon and a high-resolution lithography mask. We further implement a PnP reconstruction algorithm with cascaded denoisers for high quality reconstruction. Based on the proposed HCA-SCI system and algorithm, we achieve a 10-mega pixel SCI system to capture high-speed scenes, leading to a high throughput of 4.6G voxels per second. Both simulation and real data experiments verify the feasibility and performance of our proposed HCA-SCI scheme.

eess.IV↗

arXiv2021

Carton dataset synthesis method for domain shift based on foreground texture decoupling and replacement

One major impediment in rapidly deploying object detection models for industrial applications is the lack of large annotated datasets. We currently have presented the Sacked Carton Dataset(SCD) that contains carton images from three scenarios, such as comprehensive pharmaceutical logistics company(CPLC), e-commerce logistics company(ECLC), fruit market(FM). However, due to domain shift, the model trained with one of the three scenarios in SCD has poor generalization ability when applied to the rest scenarios. To solve this problem, a novel image synthesis method is proposed to replace the foreground texture of the source datasets with the texture of the target datasets. Our method can keep the context relationship of foreground objects and backgrounds unchanged and greatly augment the target datasets. We firstly propose a surface segmentation algorithm to achieve texture decoupling of each instance. Secondly, a contour reconstruction algorithm is proposed to keep the occlusion and truncation relationship of the instance unchanged. Finally, the Gaussian fusion algorithm is used to replace the foreground texture from the source datasets with the texture from the target datasets. The novel image synthesis method can largely boost AP by at least 4.3%~6.5% on RetinaNet and 3.4%~6.8% on Faster R-CNN for the target domain. Code is available at https://github.com/hustgetlijun/RCAN.

cs.CV↗

arXiv2020

Relative wealth concerns with partial information and heterogeneous priors

We establish a Nash equilibrium in a market with $ N $ agents with the performance criteria of relative wealth level when the market return is unobservable. Each investor has a random prior belief on the return rate of the risky asset. The investors can be heterogeneous in both the mean and variance of the prior. By a separation result and a martingale argument, we show that the optimal investment strategy under a stochastic return rate model can be characterized by a fully-coupled linear FBSDE. Two sets of deep neural networks are used for the numerical computation to first find each investor's estimate of the mean return rate and then solve the FBSDEs. We establish the existence and uniqueness result for the class of FBSDEs with stochastic coefficients and solve the utility game under partial information using deep neural network function approximators. We demonstrate the efficiency and accuracy by a base-case comparison with the solution from the finite difference scheme in the linear case and apply the algorithm to the general case of nonlinear hidden variable process. Simulations of investment strategies show a herd effect that investors trade more aggressively under relativeness concerns. Statistical properties of the investment strategies and the portfolio performance, including the Sharpe ratios and the Variance Risk ratios (VRRs) are examed. We observe that the agent with the most accurate prior estimate is likely to lead the herd, and the effect of competition on heterogeneous agents varies more with market characteristics compared to the homogeneous case.

q-fin.PM↗

arXiv2020

Hardware Impairments Aware Full-Duplex NOMA Networks Over Rician Fading Channels

A cooperative full duplex (FD) non-orthogonal multiple access (NOMA) scheme over Rician fading channels is considered. To be practical, imperfect successive interference cancellation (ipSIC) and residual hardware impairments (RHIs) at transceivers are taken into account. To evaluate the performance of the considered system, the analytical approximate expressions for the outage probability (OP) and the ergodic rate (ER) of the considered system are derived, and the asymptotic performance is also explored. Simulation results manifest that the FD cooperative NOMA can improve the ergodic sum rate (ESR) of the system compared to the half duplex (HD) mode.

cs.IT↗

arXiv2020

Smart Prediction of the Complaint Hotspot Problem in Mobile Network

In mobile network, a complaint hotspot problem often affects even thousands of users' service and leads to significant economic losses and bulk complaints. In this paper, we propose an approach to predict a customer complaint based on real-time user signalling data. Through analyzing the network and user sevice procedure, 30 key data fields related to user experience have been extracted in XDR data collected from the S1 interface. Furthermore, we augment these basic features with derived features for user experience evaluation, such as one-hot features, statistical features and differential features. Considering the problems of unbalanced data, we use LightGBM as our prediction model. LightGBM has strong generalization ability and was designed to handle unbalanced data. Experiments we conducted prove the effectiveness and efficiency of this proposal. This approach has been deployed for daily routine to locate the hot complaint problem scope as well as to report affected users and area.

cs.NI↗

arXiv2019

Sogou Machine Reading Comprehension Toolkit

Machine reading comprehension have been intensively studied in recent years, and neural network-based models have shown dominant performances. In this paper, we present a Sogou Machine Reading Comprehension (SMRC) toolkit that can be used to provide the fast and efficient development of modern machine comprehension models, including both published models and original prototypes. To achieve this goal, the toolkit provides dataset readers, a flexible preprocessing pipeline, necessary neural network components, and built-in models, which make the whole process of data preparation, model construction, and training easier.

cs.CL↗

arXiv2018

High fidelity single-pixel imaging

Single-pixel imaging (SPI) is an emerging technique which has attracts wide attention in various research fields. However, restricted by the low reconstruction quality and large amount of measurements, the practical application is still in its infancy. Inspired by the fact that natural scenes exhibit unique degenerate structures in the low dimensional subspace, we propose to take advantage of the local prior in convolutional sparse coding to implement high fidelity single-pixel imaging. Specifically, by statistically learning strategy, the target scene can be sparse represented on an overcomplete dictionary. The dictionary is composed of various basis learned from a natural image database. We introduce the above local prior into conventional SPI framework to promote the final reconstruction quality. Experiments both on synthetic data and real captured data demonstrate that our method can achieve better reconstruction from the same measurements, and thus consequently reduce the number of required measurements for same reconstruction quality.

eess.IV↗

arXiv2018

Snapshot hyperspectral imaging via spectral basis multiplexing in Fourier domain

Hyperspectral imaging is an important tool having been applied in various fields, but still limited in observation of dynamic scenes. In this paper, we propose a snapshot hyperspectral imaging technique which exploits both spectral and spatial sparsity of natural scenes. Under the computational imaging scheme, we conduct spectral dimension reduction and spatial frequency truncation to the hyperspectral data cube and snapshot it in a low cost manner. Specifically, we modulate the spectral variations by several broadband spectral filters, and then map these modulated images into different regions in the Fourier domain. The encoded image compressed in both spectral and spatial are finally collected by a monochrome detector. Correspondingly, the reconstruction is essentially a Fourier domain extraction and spectral dimensional back projection with low computational load. This Fourier-spectral multiplexing in a 2D sensor simplifies both the encoding and decoding process, and makes hyperspectral data captured in a low cost manner. We demonstrate the high performance of our method by quantitative evaluation on simulation data and build a prototype system experimentally for further validation.

physics.ins-det↗

arXiv2018

Identifying viruses from metagenomic data by deep learning

The recent development of metagenomic sequencing makes it possible to sequence microbial genomes including viruses in an environmental sample. Identifying viral sequences from metagenomic data is critical for downstream virus analyses. The existing reference-based and gene homology-based methods are not efficient in identifying unknown viruses or short viral sequences. Here we have developed a reference-free and alignment-free machine learning method, DeepVirFinder, for predicting viral sequences in metagenomic data using deep learning techniques. DeepVirFinder was trained based on a large number of viral sequences discovered before May 2015. Evaluated on the sequences after that date, DeepVirFinder outperformed the state-of-the-art method VirFinder at all contig lengths. Enlarging the training data by adding millions of purified viral sequences from environmental metavirome samples significantly improves the accuracy for predicting under-represented viruses. Applying DeepVirFinder to real human gut metagenomic samples from patients with colorectal carcinoma (CRC) identified 51,138 viral sequences belonging to 175 bins. Ten bins were associated with the cancer status, indicating their potential use for non-invasive diagnosis of CRC. In summary, DeepVirFinder greatly improved the precision and recall rates of viral identification, and it will significantly accelerate the discovery rate of viruses.

q-bio.GN↗

arXiv2018

Estimating the number of species to attain sufficient representation in a random sample

The statistical problem of using an initial sample to estimate the number of species in a larger sample has found important applications in fields far removed from ecology. Here we address the general problem of estimating the number of species that will be represented by at least a number r of observations in a future sample. The number r indicates species with sufficient observations, which are commonly used as a necessary condition for any robust statistical inference. We derive a procedure to construct consistent estimators that apply universally for a given population: once constructed, they can be evaluated as a simple function of r. Our approach is based on a relation between the number of species represented at least r times and the higher derivatives of the expected number of species discovered per unit of time. Combining this relation with a rational function approximation, we propose nonparametric estimators that are accurate for both large values of r and long-range extrapolations. We further show that our estimators retain asymptotic behaviors that are essential for applications on large-scale datasets. We evaluate the performance of this approach by both simulation and real data applications for inferences of the vocabulary of Shakespeare and Dickens, the topology of a Twitter social network, and molecular diversity in DNA sequencing data.

stat.ME↗