arXiv subjects

Bin Dai

Publications and source records attributed to Bin Dai.

At least 55 records · Page 3Linked to original sources

arXiv2021

Attentional Graph Neural Network for Parking-slot Detection

Deep learning has recently demonstrated its promising performance for vision-based parking-slot detection. However, very few existing methods explicitly take into account learning the link information of the marking-points, resulting in complex post-processing and erroneous detection. In this paper, we propose an attentional graph neural network based parking-slot detection method, which refers the marking-points in an around-view image as graph-structured data and utilize graph neural network to aggregate the neighboring information between marking-points. Without any manually designed post-processing, the proposed method is end-to-end trainable. Extensive experiments have been conducted on public benchmark dataset, where the proposed method achieves state-of-the-art accuracy. Code is publicly available at \url{https://github.com/Jiaolong/gcn-parking-slot}.

cs.CV↗

arXiv2021

Active Disturbance Rejection Control Design with Suppression of Sensor Noise Effects in Application to DC-DC Buck Power Converter

The performance of active disturbance rejection control (ADRC) algorithms can be limited in practice by high-frequency measurement noise. In this work, this problem is addressed by transforming the high-gain extended state observer (ESO), which is the inherent element of ADRC, into a new cascade observer structure. Set of experiments, performed on a DC-DC buck power converter system, show that the new cascade ESO design, compared to the conventional approach, effectively suppresses the detrimental effect of sensor noise over-amplification while increasing the estimation/control performance. The proposed design is also analyzed with a low-pass filter at the converter output, which is a common technique for reducing measurement noise in industrial applications.

eess.SY↗

arXiv2020

Further Analysis of Outlier Detection with Deep Generative Models

The recent, counter-intuitive discovery that deep generative models (DGMs) can frequently assign a higher likelihood to outliers has implications for both outlier detection applications as well as our overall understanding of generative modeling. In this work, we present a possible explanation for this phenomenon, starting from the observation that a model's typical set and high-density region may not conincide. From this vantage point we propose a novel outlier test, the empirical success of which suggests that the failure of existing likelihood-based outlier tests does not necessarily imply that the corresponding generative model is uncalibrated. We also conduct additional experiments to help disentangle the impact of low-level texture versus high-level semantics in differentiating outliers. In aggregate, these results suggest that modifications to the standard evaluation practices and benchmarks commonly applied in the literature are needed.

stat.ML↗

arXiv2020

Drosophila-Inspired 3D Moving Object Detection Based on Point Clouds

3D moving object detection is one of the most critical tasks in dynamic scene analysis. In this paper, we propose a novel Drosophila-inspired 3D moving object detection method using Lidar sensors. According to the theory of elementary motion detector, we have developed a motion detector based on the shallow visual neural pathway of Drosophila. This detector is sensitive to the movement of objects and can well suppress background noise. Designing neural circuits with different connection modes, the approach searches for motion areas in a coarse-to-fine fashion and extracts point clouds of each motion area to form moving object proposals. An improved 3D object detection network is then used to estimate the point clouds of each proposal and efficiently generates the 3D bounding boxes and the object categories. We evaluate the proposed approach on the widely-used KITTI benchmark, and state-of-the-art performance was obtained by using the proposed approach on the task of motion detection.

cs.CV↗

arXiv2019

The Usual Suspects? Reassessing Blame for VAE Posterior Collapse

In narrow asymptotic settings Gaussian VAE models of continuous data have been shown to possess global optima aligned with ground-truth distributions. Even so, it is well known that poor solutions whereby the latent posterior collapses to an uninformative prior are sometimes obtained in practice. However, contrary to conventional wisdom that largely assigns blame for this phenomena on the undue influence of KL-divergence regularization, we will argue that posterior collapse is, at least in part, a direct consequence of bad local minima inherent to the loss surface of deep autoencoder networks. In particular, we prove that even small nonlinear perturbations of affine VAE decoder models can produce such minima, and in deeper models, analogous minima can force the VAE to behave like an aggressive truncation operator, provably discarding information along all latent dimensions in certain circumstances. Regardless, the underlying message here is not meant to undercut valuable existing explanations of posterior collapse, but rather, to refine the discussion and elucidate alternative risk factors that may have been previously underappreciated.

cs.LG↗

arXiv2019

Diagnosing and Enhancing VAE Models

Although variational autoencoders (VAEs) represent a widely influential deep generative model, many aspects of the underlying energy function remain poorly understood. In particular, it is commonly believed that Gaussian encoder/decoder assumptions reduce the effectiveness of VAEs in generating realistic samples. In this regard, we rigorously analyze the VAE objective, differentiating situations where this belief is and is not actually true. We then leverage the corresponding insights to develop a simple VAE enhancement that requires no additional hyperparameters or sensitive tuning. Quantitatively, this proposal produces crisp samples and stable FID scores that are actually competitive with a variety of GAN models, all while retaining desirable attributes of the original VAE architecture. A shorter version of this work will appear in the ICLR 2019 conference proceedings (Dai and Wipf, 2019). The code for our model is available at https://github.com/daib13/ TwoStageVAE.

cs.LG↗

arXiv2019

Hidden Talents of the Variational Autoencoder

Variational autoencoders (VAE) represent a popular, flexible form of deep generative model that can be stochastically fit to samples from a given random process using an information-theoretic variational bound on the true underlying distribution. Once so-obtained, the model can be putatively used to generate new samples from this distribution, or to provide a low-dimensional latent representation of existing samples. While quite effective in numerous application domains, certain important mechanisms which govern the behavior of the VAE are obfuscated by the intractable integrals and resulting stochastic approximations involved. Moreover, as a highly non-convex model, it remains unclear exactly how minima of the underlying energy relate to original design purposes. We attempt to better quantify these issues by analyzing a series of tractable special cases of increasing complexity. In doing so, we unveil interesting connections with more traditional dimensionality reduction models, as well as an intrinsic yet underappreciated propensity for robustly dismissing sparse outliers when estimating latent manifolds. With respect to the latter, we demonstrate that the VAE can be viewed as the natural evolution of recent robust PCA models, capable of learning nonlinear manifolds of unknown dimension obscured by gross corruptions.

cs.LG↗

arXiv2019

Lidar-based Object Classification with Explicit Occlusion Modeling

LIDAR is one of the most important sensors for Unmanned Ground Vehicles (UGV). Object detection and classification based on lidar point cloud is a key technology for UGV. In object detection and classification, the mutual occlusion between neighboring objects is an important factor affecting the accuracy. In this paper, we consider occlusion as an intrinsic property of the point cloud data. We propose a novel approach that explicitly model the occlusion. The occlusion property is then taken into account in the subsequent classification step. We perform experiments on the KITTI dataset. Experimental results indicate that by utilizing the occlusion property that we modeled, the classifier obtains much better performance.

cs.RO↗

arXiv2018

An Improved Feedback Coding Scheme for the Wire-tap Channel

The model of wiretap channel (WTC) is important as it constitutes the essence of physical layer security (PLS). Wiretap channel with noiseless feedback (WTC-NF) is especially interesting as it shows what can be done when a private feedback is available. The already existing secret key based feedback coding scheme focuses on generating key from the feedback and using this key to protect part of the transmitted message. It has been shown that this secret key based feedback coding scheme is only optimal for the degraded WTC-NF, and finding an optimal feedback scheme for the general WTC-NF motivates us to exploit other uses of the feedback. In this paper, a new feedback coding scheme for the general WTC-NF is proposed, where the feedback is not only used to generate key, but also used to generate help information which helps the legitimate parties to improve the communication between them. We show that the proposed new feedback scheme performs better than the already existing one, and a binary example is given to further explain the results of this paper.

cs.IT↗

arXiv2018

Compressing Neural Networks using the Variational Information Bottleneck

Neural networks can be compressed to reduce memory and computational requirements, or to increase accuracy by facilitating the use of a larger base architecture. In this paper we focus on pruning individual neurons, which can simultaneously trim model size, FLOPs, and run-time memory. To improve upon the performance of existing compression algorithms we utilize the information bottleneck principle instantiated via a tractable variational bound. Minimization of this information theoretic bound reduces the redundancy between adjacent layers by aggregating useful information into a subset of neurons that can be preserved. In contrast, the activations of disposable neurons are shut off via an attractive form of sparse regularization that emerges naturally from this framework, providing tangible advantages over traditional sparsity penalties without contributing additional tuning parameters to the energy landscape. We demonstrate state-of-the-art compression rates across an array of datasets and network architectures.

cs.CV↗

arXiv2018

Feedback Coding Schemes for the Broadcast Channel with Mutual Secrecy Requirement

Recently, the physical layer security (PLS) of the communication systems has been shown to be enhanced by using legal receiver's feedback. The present secret key based feedback scheme mainly focuses on producing key from the feedback and using this key to protect part of the transmitted message. However, this feedback scheme has been proved only optimal for several degraded cases. The broadcast channel with mutual secrecy requirement (BC-MSR) is important as it constitutes the essence of physical layer security (PLS) in the down-link of the wireless communication systems. In this paper, we investigate the feedback effects on the BC-MSR, and propose two inner bounds and one outer bound on the secrecy capacity region of the BC-MSR with noiseless feedback. One inner bound is constructed according to the already existing secret key based feedback coding scheme for the wiretap channel, and the other is constructed by a hybrid coding scheme using feedback to generate not only keys protecting the transmitted messages but also cooperative messages helping the receivers to improve their decoding performance. The performance of the proposed feedback schemes and the gap between the inner and outer bounds are further explained via two examples.

cs.IT↗

arXiv2018

Finite State Multiple-Access Wiretap Channel with Delayed Feedback

Recently, it has been shown that the time-varying multiple-access channel (MAC) with perfect channel state information (CSI) at the receiver and delayed feedback CSI at the transmitters can be modeled as the finite state MAC (FS-MAC) with delayed state feedback, where the time variation of the channel is characterized by the statistics of the underlying state process. To study the fundamental limit of the secure transmission over multi-user wireless communication systems, we re-visit the FS-MAC with delayed state feedback by considering an external eavesdropper, which we call the finite state multiple-access wiretap channel (FS-MAC-WT) with delayed feedback. The main contribution of this paper is to show that taking full advantage of the delayed channel output feedback helps to increase the secrecy rate region of the FS-MAC-WT with delayed state feedback, and the results of this paper are further illustrated by a degraded Gaussian fading example.

cs.IT↗

arXiv2017

Design of Polar Codes with Single and Multi-Carrier Modulation on Impulsive Noise Channels using Density Evolution

In this paper, density evolution-based construction methods to design good polar codes on impulsive noise channels for single-carrier and multi-carrier systems are proposed and evaluated. For a single-carrier system, the tight bound of the block error probability (BLEP) is derived by applying density evolution and the performance of the proposed construction methods are compared. For the multi-carrier system employing orthogonal frequency-division multiplexing, the accurate BLEP estimation is not feasible so a tight lower bound on the BLEP for polar codes is derived by assuming the noise on each sub-carrier is Gaussian. The results show that the lower bound becomes tighter as the number of carriers increases.

cs.IT↗

arXiv2017

Understanding and Predicting The Attractiveness of Human Action Shot

Selecting attractive photos from a human action shot sequence is quite challenging, because of the subjective nature of the "attractiveness", which is mainly a combined factor of human pose in action and the background. Prior works have actively studied high-level image attributes including interestingness, memorability, popularity, and aesthetics. However, none of them has ever studied the "attractiveness" of human action shot. In this paper, we present the first study of the "attractiveness" of human action shots by taking a systematic data-driven approach. Specifically, we create a new action-shot dataset composed of about 8000 high quality action-shot photos. We further conduct rich crowd-sourced human judge studies on Amazon Mechanical Turk(AMT) in terms of global attractiveness of a single photo, and relative attractiveness of a pair of photos. A deep Siamese network with a novel hybrid distribution matching loss was further proposed to fully exploit both types of ratings. Extensive experiments reveal that (1) the property of action shot attractiveness is subjective but predicable (2) our proposed method is both efficient and effective for predicting the attractive human action shots.

cs.CV↗

arXiv2017

Achieving both positive secrecy rates of the users in two-way wiretap channel by individual secrecy

In this paper, the individual secrecy of two-way wiretap channel is investigated, where two legitimate users' messages are separately guaranteed secure against an external eavesdropper. For one thing, in some communication scenarios, the joint secrecy is impossible to achieve both positive secrecy rates of two users. For another, the individual secrecy satisfies the secrecy demand of many practical communication systems. Thus, firstly, an achievable secrecy rate region is derived for the general two-way wiretap channel with individual secrecy. In a deterministic channel, the region with individual secrecy is shown to be larger than that with joint secrecy.Secondly, outer bounds on the secrecy capacity region are obtained for the general two-way wiretap channel and for two classes of special two-way wiretap channels.The gap between inner and outer bounds on the secrecy capacity region is explored via the binary input two-way wiretap channels and the degraded Gaussian two-way wiretap. Most notably, the secrecy capacity regions are established for the XOR channel and the degraded Gaussian two-way wiretap channel.Furthermore, the secure sum-rate of the degraded Gaussian two-way wiretap channel under the individual secrecy constraint is demonstrated to be strictly larger than that under the joint secrecy constraint.

cs.IT↗

arXiv2017

Multiple Access Wiretap Channel with Noiseless Feedback

The physical layer security in the up-link of the wireless communication systems is often modeled as the multiple access wiretap channel (MAC-WT), and recently it has received a lot attention. In this paper, the MAC-WT has been re-visited by considering the situation that the legitimate receiver feeds his received channel output back to the transmitters via two noiseless channels, respectively. This model is called the MAC-WT with noiseless feedback. Inner and outer bounds on the secrecy capacity region of this feedback model are provided. To be specific, we first present a decode-and-forward (DF) inner bound on the secrecy capacity region of this feedback model, and this bound is constructed by allowing each transmitter to decode the other one's transmitted message from the feedback, and then each transmitter uses the decoded message to re-encode his own messages, i.e., this DF inner bound allows the independent transmitters to co-operate with each other. Then, we provide a hybrid inner bound which is strictly larger than the DF inner bound, and it is constructed by using the feedback as a tool not only to allow the independent transmitters to co-operate with each other, but also to generate two secret keys respectively shared between the legitimate receiver and the two transmitters. Finally, we give a sato-type outer bound on the secrecy capacity region of this feedback model. The results of this paper are further explained via a Gaussian example.

cs.IT↗

arXiv2017

Finite State Markov Wiretap Channel with Delayed Feedback

The finite state Markov channel (FSMC), where the channel transition probability is controlled by a state undergoing a Markov process, is a useful model for the mobile wireless communication channel. In this paper, we investigate the security issue in the mobile wireless communication systems by considering the FSMC with an eavesdropper, which we call the finite state Markov wiretap channel (FSM-WC). We assume that the state is perfectly known by the legitimate receiver and the eavesdropper, and through a noiseless feedback channel, the legitimate receiver sends his received channel output and the state back to the transmitter after some time delay. Inner and outer bounds on the capacity-equivocation regions of the FSM-WC with delayed state feedback and with or without delayed channel output feedback are provided in this paper, and we show that these bounds meet if the eavesdropper's received symbol is a degraded version of the legitimate receiver's. The above results are further explained via degraded Gaussian and Gaussian fading examples.

cs.IT↗

arXiv2015

Relay Broadcast Channel with Confidential Messages

We investigate the effects of an additional relay node on the secrecy of broadcast channels by considering the model of relay broadcast channels with confidential messages. We show that this additional relay node can increase the achievable secrecy rate region of the broadcast channels with confidential messages. More specifically, first, we investigate the discrete memoryless relay broadcast channels with two confidential messages and one common message. Three inner bounds (with respect to decode-forward, generalized noise-forward and compress-forward strategies) and an outer bound on the capacity-equivocation region are provided. Second, we investigate the discrete memoryless relay broadcast channels with two confidential messages. Inner and outer bounds on the capacity-equivocation region are provided. Finally, we investigate the discrete memoryless relay broadcast channels with one confidential message and one common message. Inner and outer bounds on the capacity-equivocation region are provided, and the results are further explained via a Gaussian example. Compared with Csiszar-Korner's work on broadcast channels with confidential messages (BCC), we find that with the help of the relay node, the secrecy capacity region of the Gaussian BCC is enhanced.

cs.IT↗