The Outstanding performance and growing size of Large Language Models has led to increased attention in parameter efficient learning. The two predominant approaches are Adapters and Pruning. Adapters are to freeze the model and give it a new weight matrix on the side, which can significantly reduce the time and memory of training, but the cost is that the evaluation and testing will increase the time and memory consumption. Pruning is to cut off some weight and re-distribute the remaining weight, which sacrifices the complexity of training at the cost of extremely high memory and training time, making the cost of evaluation and testing relatively low. So efficiency of training and inference can't be obtained in the same time. In this work, we propose a task-oriented Pruning-Adapter method that achieve a high memory efficiency of training and memory, and speeds up training time and ensures no significant decrease in accuracy in GLUE tasks, achieving training and inference efficiency at the same time.
Robotic eye-in-hand calibration is the task of determining the rigid 6-DoF pose of the camera with respect to the robot end-effector frame. In this paper, we formulate this task as a non-linear optimization problem and introduce an active vision approach to strategically select the robot pose for maximizing calibration accuracy. Specifically, given an initial collection of measurement sets, our system first computes the calibration parameters and estimates the parameter uncertainties. We then predict the next robot pose from which to collect the next measurement that brings about the maximum information gain (uncertainty reduction) in the calibration parameters. We test our approach on a simulated dataset and validate the results on a real 6-axis robot manipulator. The results demonstrate that our approach can achieve accurate calibrations using many fewer viewpoints than other commonly used baseline calibration methods.
RGB-D semantic segmentation can be advanced with convolutional neural networks due to the availability of Depth data. Although objects cannot be easily discriminated by just the 2D appearance, with the local pixel difference and geometric patterns in Depth, they can be well separated in some cases. Considering the fixed grid kernel structure, CNNs are limited to lack the ability to capture detailed, fine-grained information and thus cannot achieve accurate pixel-level semantic segmentation. To solve this problem, we propose a Pixel Difference Convolutional Network (PDCNet) to capture detailed intrinsic patterns by aggregating both intensity and gradient information in the local range for Depth data and global range for RGB data, respectively. Precisely, PDCNet consists of a Depth branch and an RGB branch. For the Depth branch, we propose a Pixel Difference Convolution (PDC) to consider local and detailed geometric information in Depth data via aggregating both intensity and gradient information. For the RGB branch, we contribute a lightweight Cascade Large Kernel (CLK) to extend PDC, namely CPDC, to enjoy global contexts for RGB data and further boost performance. Consequently, both modal data's local and global pixel differences are seamlessly incorporated into PDCNet during the information propagation process. Experiments on two challenging benchmark datasets, i.e., NYUDv2 and SUN RGB-D reveal that our PDCNet achieves state-of-the-art performance for the semantic segmentation task.
This paper studies the exploitation of triple polarization (TP) for multi-user (MU) holographic multiple-input multiple-output surface (HMIMOS) wireless communication systems, aiming at capacity boosting without enlarging the antenna array size. We specifically consider that both the transmitter and receiver are equipped with an HMIMOS comprising compact sub-wavelength TP patch antennas. To characterize TP MUHMIMOS systems, a TP near-field channel model is proposed using the dyadic Green's function, whose characteristics are leveraged to design a user-cluster-based precoding scheme for mitigating the cross-polarization and inter-user interference contributions. A theoretical correlation analysis for HMIMOS with infinitely small patch antennas is also presented. According to the proposed scheme, the users are assigned to one of the three polarizations, which is easy to implement, at the cost, however, of reducing the system's diversity. Our numerical results showcase that the cross-polarization channel components have a nonnegligible impact on the system performance, which is efficiently eliminated with the proposed MU precoding scheme.
In recent years, arbitrary image style transfer has attracted more and more attention. Given a pair of content and style images, a stylized one is hoped that retains the content from the former while catching style patterns from the latter. However, it is difficult to simultaneously keep well the trade-off between the content details and the style features. To stylize the image with sufficient style patterns, the content details may be damaged and sometimes the objects of images can not be distinguished clearly. For this reason, we present a new transformer-based method named STT for image style transfer and an edge loss which can enhance the content details apparently to avoid generating blurred results for excessive rendering on style features. Qualitative and quantitative experiments demonstrate that STT achieves comparable performance to state-of-the-art image style transfer methods while alleviating the content leak problem.
Offline reinforcement learning (RL) enables the agent to effectively learn from logged data, which significantly extends the applicability of RL algorithms in real-world scenarios where exploration can be expensive or unsafe. Previous works have shown that extracting primitive skills from the recurring and temporally extended structures in the logged data yields better learning. However, these methods suffer greatly when the primitives have limited representation ability to recover the original policy space, especially in offline settings. In this paper, we give a quantitative characterization of the performance of offline hierarchical learning and highlight the importance of learning lossless primitives. To this end, we propose to use a \emph{flow}-based structure as the representation for low-level policies. This allows us to represent the behaviors in the dataset faithfully while keeping the expression ability to recover the whole policy space. We show that such lossless primitives can drastically improve the performance of hierarchical policies. The experimental results and extensive ablation studies on the standard D4RL benchmark show that our method has a good representation ability for policies and achieves superior performance in most tasks.
With the breakthrough of AlphaGo, deep reinforcement learning becomes a recognized technique for solving sequential decision-making problems. Despite its reputation, data inefficiency caused by its trial and error learning mechanism makes deep reinforcement learning hard to be practical in a wide range of areas. Plenty of methods have been developed for sample efficient deep reinforcement learning, such as environment modeling, experience transfer, and distributed modifications, amongst which, distributed deep reinforcement learning has shown its potential in various applications, such as human-computer gaming, and intelligent transportation. In this paper, we conclude the state of this exciting field, by comparing the classical distributed deep reinforcement learning methods, and studying important components to achieve efficient distributed learning, covering single player single agent distributed deep reinforcement learning to the most complex multiple players multiple agents distributed deep reinforcement learning. Furthermore, we review recently released toolboxes that help to realize distributed deep reinforcement learning without many modifications of their non-distributed versions. By analyzing their strengths and weaknesses, a multi-player multi-agent distributed deep reinforcement learning toolbox is developed and released, which is further validated on Wargame, a complex environment, showing usability of the proposed toolbox for multiple players and multiple agents distributed deep reinforcement learning under complex games. Finally, we try to point out challenges and future trends, hoping this brief review can provide a guide or a spark for researchers who are interested in distributed deep reinforcement learning.
This paper investigates the utilization of triple polarization (TP) for multi-user (MU) holographic multiple-input multi-output surface (HMIMOS) wireless communication systems, targeting capacity boosting and diversity exploitation without enlarging the antenna array sizes. We specifically consider that both the transmitter and receiver are both equipped with an HMIMOS consisting of compact sub-wavelength TP patch antennas within the near-field (NF) regime. To characterize TP MU-HMIMOS systems, a TP NF channel model is constructed using the dyadic Green's function, whose characteristics are leveraged to design two precoding schemes for mitigating the cross-polarization and inter-user interference contributions. Specifically, a user-cluster-based precoding scheme assigns different users to one of three polarizations at the expense of the system's diversity, and a two-layer precoding scheme removes interference using the Gaussian elimination method at a high computational cost. The theoretical correlation analysis for HMIMOS in the NF region is also investigated, revealing that both the spacing of transmit patch antennas and user distance impact transmit correlation factors. Our numerical results show that the users far from transmitting HMIMOS experience higher correlation than those closer within the NF regime, resulting in a lower channel capacity. Meanwhile, in terms of channel capacity, TP HMIMOS can almost achieve 1.25 times gain compared with dual-polarized HMIMOS, and 3 times compared with conventional HMIMOS. In addition, the proposed two-layer precoding scheme combined with two-layer power allocation realizes a higher spectral efficiency than other schemes without sacrificing diversity.
Combining RGB images and the corresponding depth maps in semantic segmentation proves the effectiveness in the past few years. Existing RGB-D modal fusion methods either lack the non-linear feature fusion ability or treat both modal images equally, regardless of the intrinsic distribution gap or information loss. Here we find that depth maps are suitable to provide intrinsic fine-grained patterns of objects due to their local depth continuity, while RGB images effectively provide a global view. Based on this, we propose a pixel differential convolution attention (DCA) module to consider geometric information and local-range correlations for depth data. Furthermore, we extend DCA to ensemble differential convolution attention (EDCA) which propagates long-range contextual dependencies and seamlessly incorporates spatial distribution for RGB data. DCA and EDCA dynamically adjust convolutional weights by pixel difference to enable self-adaptive in local and long range, respectively. A two-branch network built with DCA and EDCA, called Differential Convolutional Network (DCANet), is proposed to fuse local and global information of two-modal data. Consequently, the individual advantage of RGB and depth data are emphasized. Our DCANet is shown to set a new state-of-the-art performance for RGB-D semantic segmentation on two challenging benchmark datasets, i.e., NYUDv2 and SUN-RGBD.
The current fusion positioning systems are mainly based on filtering algorithms, such as Kalman filtering or particle filtering. However, the system complexity of practical application scenarios is often very high, such as noise modeling in pedestrian inertial navigation systems, or environmental noise modeling in fingerprint matching and localization algorithms. To solve this problem, this paper proposes a fusion positioning system based on deep learning and proposes a transfer learning strategy for improving the performance of neural network models for samples with different distributions. The results show that in the whole floor scenario, the average positioning accuracy of the fusion network is 0.506m. The experiment results of transfer learning show that the estimation accuracy of the inertial navigation positioning step size and rotation angle of different pedestrians can be improved by 53.3% on average, the Bluetooth positioning accuracy of different devices can be improved by 33.4%, and the fusion can be improved by 31.6%.