arXiv subjects

Zhiyuan Wu

Publications and source records attributed to Zhiyuan Wu.

41 records · Page 3Linked to original sources

arXiv2022

Activation Map Adaptation for Effective Knowledge Distillation

Model compression becomes a recent trend due to the requirement of deploying neural networks on embedded and mobile devices. Hence, both accuracy and efficiency are of critical importance. To explore a balance between them, a knowledge distillation strategy is proposed for general visual representation learning. It utilizes our well-designed activation map adaptive module to replace some blocks of the teacher network, exploring the most appropriate supervisory features adaptively during the training process. Using the teacher's hidden layer output to prompt the student network to train so as to transfer effective semantic information.To verify the effectiveness of our strategy, this paper applied our method to cifar-10 dataset. Results demonstrate that the method can boost the accuracy of the student network by 0.6% with 6.5% loss reduction, and significantly improve its training speed.

cs.CV↗

arXiv2022

Meta-Learning-Based Deep Reinforcement Learning for Multiobjective Optimization Problems

Deep reinforcement learning (DRL) has recently shown its success in tackling complex combinatorial optimization problems. When these problems are extended to multiobjective ones, it becomes difficult for the existing DRL approaches to flexibly and efficiently deal with multiple subproblems determined by weight decomposition of objectives. This paper proposes a concise meta-learning-based DRL approach. It first trains a meta-model by meta-learning. The meta-model is fine-tuned with a few update steps to derive submodels for the corresponding subproblems. The Pareto front is then built accordingly. Compared with other learning-based methods, our method can greatly shorten the training time of multiple submodels. Due to the rapid and excellent adaptability of the meta-model, more submodels can be derived so as to increase the quality and diversity of the found solutions. The computational experiments on multiobjective traveling salesman problems and multiobjective vehicle routing problem with time windows demonstrate the superiority of our method over most of learning-based and iteration-based approaches.

cs.AI↗

arXiv2021

Spirit Distillation: A Model Compression Method with Multi-domain Knowledge Transfer

Recent applications pose requirements of both cross-domain knowledge transfer and model compression to machine learning models due to insufficient training data and limited computational resources. In this paper, we propose a new knowledge distillation model, named Spirit Distillation (SD), which is a model compression method with multi-domain knowledge transfer. The compact student network mimics out a representation equivalent to the front part of the teacher network, through which the general knowledge can be transferred from the source domain (teacher) to the target domain (student). To further improve the robustness of the student, we extend SD to Enhanced Spirit Distillation (ESD) in exploiting a more comprehensive knowledge by introducing the proximity domain which is similar to the target domain for feature extraction. Results demonstrate that our method can boost mIOU and high-precision accuracy by 1.4% and 8.2% respectively with 78.2% segmentation variance, and can gain a precise compact network with only 41.8% FLOPs.

cs.CV↗

arXiv2021

Spirit Distillation: Precise Real-time Semantic Segmentation of Road Scenes with Insufficient Data

Semantic segmentation of road scenes is one of the key technologies for realizing autonomous driving scene perception, and the effectiveness of deep Convolutional Neural Networks(CNNs) for this task has been demonstrated. State-of-art CNNs for semantic segmentation suffer from excessive computations as well as large-scale training data requirement. Inspired by the ideas of Fine-tuning-based Transfer Learning (FTT) and feature-based knowledge distillation, we propose a new knowledge distillation method for cross-domain knowledge transference and efficient data-insufficient network training, named Spirit Distillation(SD), which allow the student network to mimic the teacher network to extract general features, so that a compact and accurate student network can be trained for real-time semantic segmentation of road scenes. Then, in order to further alleviate the trouble of insufficient data and improve the robustness of the student, an Enhanced Spirit Distillation (ESD) method is proposed, which commits to exploit a more comprehensive general features extraction capability by considering images from both the target and the proximity domains as input. To our knowledge, this paper is a pioneering work on the application of knowledge distillation to few-shot learning. Persuasive experiments conducted on Cityscapes semantic segmentation with the prior knowledge transferred from COCO2017 and KITTI demonstrate that our methods can train a better student network (mIOU and high-precision accuracy boost by 1.4% and 8.2% respectively, with 78.2% segmentation variance) with only 41.8% FLOPs (see Fig. 1).

cs.CV↗

arXiv2008

Capacity and Performance of Adaptive MIMO System Based on Beam-Nulling

In this paper, we propose a scheme called "beam-nulling" for MIMO adaptation. In the beam-nulling scheme, the eigenvector of the weakest subchannel is fed back and then signals are sent over a generated subspace orthogonal to the weakest subchannel. Theoretical analysis and numerical results show that the capacity of beam-nulling is closed to the optimal water-filling at medium SNR. Additionally, signal-to-interference-plus-noise ratio (SINR) of MMSE receiver is derived for beam-nulling. Then the paper presents the associated average bit-error rate (BER) of beam-nulling numerically which is verified by simulation. Simulation results are also provided to compare beam-nulling with beamforming. To improve performance further, beam-nulling is concatenated with linear dispersion code. Simulation results are also provided to compare the concatenated beam-nulling scheme with the beamforming scheme at the same data rate. Additionally, the existing beamforming and new proposed beam-nulling can be extended if more than one eigenvector is available at the transmitter. The new extended schemes are called multi-dimensional (MD) beamforming and MD beam-nulling. Theoretical analysis and numerical results in terms of capacity are also provided to evaluate the new extended schemes. Simulation results show that the MD scheme with LDC can outperform the MD scheme with STBC significantly when the data rate is high.

cs.IT↗