检索结果-内蒙古大学图书馆

Combining various training and adaptation algorithms for ensemble few-shot classification

NEURAL NETWORKS 2025年 185卷 107211页

作者： Jiang, Zhen Tang, Na Sun, Jianlong Zhan, Yongzhao Jiangsu Univ Sch Comp Sci & Commun Engn Zhenjiang Peoples R China

To mitigate the shortage of labeled data, Few-Shot Classification (FSC) methods train deep neural networks (DNNs) on a base dataset with sufficient labeled data, and then adapt them to target tasks using a few labeled data. Despite notable progress, a single FSC model remains prone to high variance and low confidence. As a result, ensemble FSC has garnered increasing attention. However, the limited labeled data and the high computational cost associated with DNNs present significant challenges for ensemble FSC methods. This paper presents a novel ensemble method that generates multiple FSC models via combining various training and adaptation algorithms. Due to the reuse of training phases, the proposed method significantly reduces the learning cost while generating base models with greater diversity. To further minimize reliance on labeled data, we provide each model with pseudo-labeled data selected by the majority vote of other models. Compared with self-training style methods, this "one-vs-others" learning strategy effectively reduces pseudo-label noise and confirmation bias. Finally, we conduct extensive experiments on miniImageNet, tieredImageNet and CUB datasets. The experimental results demonstrate that our method outperforms other state-of-the-art FSC methods. Especially, our method achieves the greatest improvement in the performance of base models. The source code and related models are available at https://***/tn1999tn/Ensemble-FSC/tree/master.

关键词： Few-shot classification Ensemble learning Training algorithms adaptation algorithms Pseudo-labeled data

来源：评论

学校读者我要写书评

暂无评论

Constrained adaptation algorithms employing householder transformation

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2002年第9期50卷 2187-2195页

作者： de Campos, MLR Werner, S Apolinário, JA Univ Fed Rio de Janeiro COPPE Programa Engn Eletr BR-21945 Rio De Janeiro Brazil Aalto Univ Signal Proc Lab FIN-02150 Espoo Finland Inst Mil Engn Dept Engn Eletr Rio De Janeiro Brazil

This paper presents a tutorial-like detailed explanation of linearly constrained minimum-variance filtering in order to introduce an efficient implementation that utilizes Householder transformation (HT). Through a graphical description of the algorithms, further insight on linearly constrained adaptive filters was made possible, and the main differences among several algorithms were highlighted. The method proposed herein, based on HT, allows direct application of any unconstrained adaptation algorithm as in a generalized sidelobe canceller (GSC), but unlike the GSC, the HT-based approach always renders efficient implementations. A complete and detailed comparison with the GSC model and a thorough discussion of the advantages of the HT-based approach are also given. Simulations were run in a beamforming application where a linear array of 12 sensors was used. It was verified that not only the HT approach yields efficient implementation of constrained adaptive filters, but in addition, the beampatterns achieved with this method were much closer to the optimal solution than the beampatterns obtained with GSC models with similar computational complexity.

关键词： adaptation algorithms beamforming efficient alogorithms generalized sidelobe canceller linearly constrained adaptive filters

来源：评论

学校读者我要写书评

暂无评论

Training fuzzy logic controller software components by combining adaptation algorithms

引用

ADVANCES IN ENGINEERING SOFTWARE 2003年第3期34卷 125-137页

作者： Chen, J Rine, DC George Mason Univ Sch Informat Technol & Engn Fairfax VA 22030 USA

An approach for an effective and efficient off-line training of particular classes of reusable controller software components is presented. To build a necessary relationship between a component's abstract and concrete levels, each control software component is represented at the abstract level by means of a set of adaptive fuzzy logic rules and at the concrete level by means of adaptive fuzzy membership functions. Training includes two phases: testing and adapting. The testing phase is for identifying faulty fuzzy elements of a component, while the adapting phase is for modifying membership functions. We employ genetic algorithms, neural network algorithms, Monte Carlo algorithms, and their combinations in each phase. This approach is illustrated by training automotive controller software components (simulation). Experimental simulation results show that our off-line training approach supports controller software component adaptation effectively and efficiently in terms of controlled process operation accuracy and effort spent. (C) 2003 Elsevier Science Ltd. All rights reserved.

关键词： software adaptation and maintenance adaptive fuzzy logic controller adaptation algorithms

来源：评论

学校读者我要写书评

暂无评论

On a general structure for adaptation/learning algorithms.-Stability and performance issues

引用

AUTOMATICA 2023年第1期156卷

作者： Landau, Ioan Dore Airimitoaie, Tudor-Bogdan Vau, Bernard Buche, Gabriel Univ Grenoble Alpes GIPSA Lab CNRS Grenoble INP F-38000 Grenoble France Univ Bordeaux CNRS Bordeaux INP IMS F-33405 Talence France EXAIL 12 Ave Coquelicots F-94385 Bonneuil Sur Marne France

The paper introduces a general structure for parameter adaptation/learning algorithms (PALA). This structure is characterized by the presence of an embedded ARMA (poles-zeros) filter in the PALA. The key question is how to select the coefficients of this filter in order, on the one hand, to guarantee the stability of the parameter estimator for any (positive) value of the adaptation gain/learning rate and for any initial conditions and on the other hand to accelerate the adaptation transient. In order to achieve this, it is shown that on one hand the embedded ARMA filter should be characterized by a positive real transfer function and on the other hand the filter acting on the correcting term (the dynamic adaptation gain) should be characterized by a strictly positive real transfer function. Specific conditions for the design of a second order ARMA embedded filter (ARIMA2 algorithm) are provided. It is shown in the paper that many parameter adaptation/learning algorithms (PALA) used in adap-tive control, system identification and neural networks (Nesterov, Conjugate gradients, Momentum back propagation, Averaged gradient, Integral+proportional+derivative, ...) are particular cases of the PALA structure introduced in this paper and specific conditions for the stable operation of these algorithms are given. Performance of the ARIMA2 algorithm as well as of the other algorithms reviewed in the paper will be comparatively evaluated by simulations and experimental results obtained on an active noise control system. & COPY;2023 Elsevier Ltd. All rights reserved.

关键词： adaptation algorithms Stability Adaptive feedforward compensation Youla-Kucera parametrization

来源：评论

学校读者我要写书评

暂无评论

adaptation engine for a streaming service based on MPEG-DASH

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2015年第18期74卷 7983-8002页

作者： Pozueco, Laura Garcia Paneda, Xabiel Garcia, Roberto Melendi, David Cabrero, Sergio Diaz Orueta, Gabriel Univ Oviedo Dept Informat Gijon Spain Spanish Univ Distance Educ UNED Elect & Comp Dept Madrid Spain

HTTP Video streaming has become a strong candidate for video transmission on the Internet thanks to the abundance of web infrastructure. With the recent standardization of the new MPEG Dynamic Adaptive Streaming over HTTP (DASH), the flexibility and implantation of adaptive video systems has increased due to the fact that DASH can operate on a conventional web infrastructure. In this paper we propose an estimation and adaptation system for DASH. The proposed adaptive algorithm is based on client buffer threshold and smooth throughput measures (based on the throughput of previous segments). Before DASH, the standard of Scalable Video Coding (SVC) also arose from the idea of adaptation. Both systems (adaptive system based on SVC and the proposed system for DASH) are compared in terms of Video Quality (VQ) metrics. The results show that the proposed system reacts properly to changes in the network capacity, while maintaining a minimum level of segments in the buffer. The user-perceived quality is better than in the SVC-based adaptive system although the generated traffic is higher.

关键词： Dynamic adaptive streaming over HTTP (DASH) Scalable video coding (SVC) adaptation algorithms Bandwidth estimation

来源：评论

学校读者我要写书评

暂无评论

Gait Pattern adaptation for an Active Lower-Limb Orthosis Based on Neural Networks

引用

ADVANCED ROBOTICS 2011年第15期25卷 1903-1925页

作者： Gomes, M. A. Silveira, G. L. M. Siqueira, A. A. G. Univ Sao Paulo Dept Mech Engn BR-13566590 Sao Carlos SP Brazil

This work deals with neural network (NN)-based gait pattern adaptation algorithms for an active lower-limb orthosis. Stable trajectories with different walking speeds are generated during an optimization process considering the zero-moment point (ZMP) criterion and the inverse dynamic of the orthosis-patient model. Additionally, a set of NNs is used to decrease the time-consuming analytical computation of the model and ZMP. The first NN approximates the inverse dynamics including the ZMP computation, while the second NN works in the optimization procedure, giving an adapted desired trajectory according to orthosis-patient interaction. This trajectory adaptation is added directly to the trajectory generator, also reproduced by a set of NNs. With this strategy, it is possible to adapt the trajectory during the walking cycle in an on-line procedure, instead of changing the trajectory parameter after each step. The dynamic model of the actual exoskeleton, with interaction forces included, is used to generate simulation results. Also, an experimental test is performed with an active ankle-foot orthosis, where the dynamic variables of this joint are replaced in the simulator by actual values provided by the device. It is shown that the final adapted trajectory follows the patient intention of increasing the walking speed, so changing the gait pattern. (C) Koninklijke Brill NV, Leiden, 2011

关键词： Exoskeleton stable gait pattern zero-moment point criterion adaptation algorithms neural network

来源：评论

学校读者我要写书评

暂无评论

Convergence of a Bakhvalov Grid adaptation Method for Singularly Perturbed Boundary Value Problems

引用

NUMERICAL ANALYSIS AND APPLICATIONS 2016年第1期9卷 34-44页

作者： Blatov, I. A. Kitaeva, E. V. Volga State Univ Telecommun & Informat Ul Lva Tolstogo 23 Samara 443010 Russia

In this paper, a Galerkin finite element method for non-self-adjoint boundary value problems on Bakhvalov grids is considered. Using the Galerkin projection method, the convergence of a sequence of computational grids... 详细信息

关键词： singularly perturbed boundary value problem Galerkin projection Bakhvalov grid adaptation algorithms

来源：评论

学校读者我要写书评

暂无评论

Energy consumption adaptation approach for Smart Buildings 14

Energy consumption adaptation approach for Smart Buildings

引用

14th IEEE/ACS International Conference on Computer Systems and Applications (AICCSA)

作者： Taktak, Emna Rodriguez, Ismael Bouassida Univ Sfax ReDCAD BP 1173 Sfax 3038 Tunisia

ISBN: (纸本)9781538635810

Since the last years, a huge interest has been paid for the new generation buildings: the Smart Buildings. The Smart Buildings bring solutions to several issues, like those related to the energy management and the occupant's comfort. Therefore, a Smart Building has the possibility to manage its energetic behaviour in order to reduce its energy consumption and ensure a healthy state and durability of its devices. We find that many research studies were conducted these last years for Smart Buildings. These research studies propose several solutions to manage the energy consumption in Smart Buildings. We are interested in research studies that not only manage the energy consumption but also guarantee that the occupant's wellbeing is maintained. We consider a Smart Building in which a Smart Building application implements a set of energy saving strategies. We propose different algorithms for the energy management embedded in the Smart Building application while keeping as possible the occupants' comfort.

关键词： Smart Building Smart Building Manager energy efficiency energy management adaptation context-aware occupant's comfort energy consumption adaptation algorithms

来源：评论

学校读者我要写书评

暂无评论

Hermitian-Based Hidden Activation Functions for adaptation of Hybrid HMM/ANN Models

Hermitian-Based Hidden Activation Functions for Adaptation o...

引用

13th Annual Conference of the International-Speech-Communication-Association

作者： Siniscalchi, Sabato Marco Li, Jinyu Lee, Chin-Hui Kore Univ Enna Fac Telemat Engn Enna Italy

ISBN: (纸本)9781622767595

This work is concerned with speaker adaptation techniques for artificial neural network (ANN) implemented as feed-forward multi-layer perceptrons (MLPs) in the context of large vocabulary continuous speech recognition (LVCSR). Most successful speaker adaptation techniques for MLPs consist of augmenting the neural architecture with a linear transformation network connected to either the input or the output layer. The weights of this additional linear layer are learned during the adaptation phase while all of the other weights are kept frozen in order to avoid over-fitting. In doing so, the structure of the speaker-dependent (SD) and speaker-independent (Si) architecture differs and the number of adaptation parameters depends upon the dimension of either the input or output layers. We propose an alternative neural architecture for speaker-adaptation to overcome the limits of current approaches. This neural architecture adopts hidden activation functions that can be learned directly from the adaptation data. This adaptive capability of the hidden activation function is achieved through the use of orthonormal Hermite polynomials. Experimental evidence gathered on the Wall Street Journal Nov92 task demonstrates the viability of the proposed technique.

关键词： Connectionist speech recognition systems Neural networks adaptation algorithms Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Improved Quasi-Newton Adaptive-Filtering Algorithm

引用

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS 2010年第8期57卷 2109-2118页

作者： Ali, Md Zulfiquar Antoniou, Andreas Univ Victoria Dept Elect & Comp Engn Victoria BC V8W 3P6 Canada

An improved quasi-Newton (QN) algorithm that performs data-selective adaptation is proposed whereby the weight vector and the inverse of the input-signal autocorrelation matrix are updated only when the a priori error exceeds a prespecified error bound. The proposed algorithm also incorporates an improved estimator of the inverse of the autocorrelation matrix. With these modifications, the proposed QN algorithm takes significantly fewer updates to converge and yields a reduced steady-state misalignment relative to a known QN algorithm proposed recently. These features of the proposed QN algorithm are demonstrated through extensive simulations. Simulations also show that the proposed QN algorithm, like the known QN algorithm, is quite robust with respect to roundoff errors introduced in fixed-point implementations.

关键词： adaptation algorithms adaptive filters convergence speed in adaptation algorithms quasi-Newton algorithms steady-state misalignment

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：