Employing the traditional least-mean-square (LMS) algorithm to estimate the weight vector of an unknown system will result in an estimation bias when the input signal of the adaptive filter is corrupted by noise. This paper proposes a bias-compensated sign algorithm (BC-SA) to address this problem. Specifically, an unbiasedness condition is employed to develop a compensation term for the classical sign algorithm (SA) to reduce the estimation bias caused by noisy inputs. The proposed BC-SA can not only reduce the estimation bias but also exhibit robustness against impulsive noise. Then the mean and mean-square performance of the BC-SA is analyzed based on Price's theorem under some frequently used statistical assumptions. Moreover, the step-size of the BC-SA is optimized based on the developed theoretical mean-square deviation (MSD). Simulation results are finally provided to evaluate the convergence performance of the proposed algorithm and to examine the theoretical findings.

The essence of improving the effect of cross-modal image–text retrieval (CIR) lies in the finer-grained modeling of homogeneous features between modalities. However, in remote sensing (RS) scenarios, existing methods usually apply the image–sentence granular feature alignment paradigm, bringing significant difficulties to the fine-grained representation of homogeneous features between modalities. Besides, more complex background noise and extreme scale ranges of foreground targets are hard to distinguish, causing the feature mottle problem. To address the above issues, we propose a novel Semantic-guided Image–text Retrieval framework with Segmentation (SIRS). It is a multitask joint learning framework for plug-and-play and end-to-end training RS CIR models efficiently, including semantic-guided spatial attention (SSA) and adaptive multiscale weighting (AMW) modules. First, SSA introduces a background reconstruction (BR) branch based on noise perception and a semantic segmentation (SS) branch based on pixel-level prediction. It explores a joint learning strategy that concisely filters background noise and refines foreground features considerably. Second, AMW performs multiscale weighting on various layers of feature map output by the encoder, effectively improving the learning efficiency of foreground targets at different scales. It is worth mentioning that SIRS outputs combination results with image and segmentation mask, which is not available in other methods. Based on the RSITMD dataset, we complete the SS annotation RSITMD-SS to verify the performance of the proposed method. Sufficient and complete experiments verify the effectiveness of the proposed method. With SIRS, the mainstream SVP and CLIP-based methods improve about 7 mR and derive segmentation prediction with acceptable computational cost optionally. The code and associated dataset will be available at https://github.com/StarBurstStream0/SIRS .

Recently, many deep learning-based methods have been developed for solving remote sensing (RS) scene classification or retrieval tasks. Most of the adopted loss functions for training these models require accurate annotations. However, the presence of noise in such annotations (also known as label noise) cannot be avoided in large-scale RS benchmark archives, resulting from geo-location/registration errors, land-cover changes, and diverse knowledge background of annotators. To overcome the influence of noisy labels on the learning process of deep models, we propose a new loss function called noise-tolerant deep neighborhood embedding which can accurately encode the semantic relationships among RS scenes. Specifically, we target at maximizing the leave-one-out K-NN score for uncovering the inherent neighborhood structure among the images in feature space. Moreover, we down-weight the contribution of potential noisy images by learning their localized structure and pruning the images with low leave-one-out K-NN scores. Based on our newly proposed loss function, classwise features can be more robustly discriminated. Our experiments, conducted on two benchmark RS datasets, validate the effectiveness of the proposed approach on three different RS scene interpretation tasks, including classification, clustering, and retrieval. The codes of this article will be publicly available from https://github.com/jiankang1991.

Endowing convolutional neural networks (CNNs) with the rotation-invariant capability is important for characterizing the semantic contents of remote sensing (RS) images since they do not have typical orientations. Most of the existing deep methods for learning rotation-invariant CNN models are based on the design of proper convolutional or pooling layers, which aims at predicting the correct category labels of the rotated RS images equivalently. However, a few works have focused on learning rotation-invariant embeddings in the framework of deep metric learning for modeling the fine-grained semantic relationships among RS images in the embedding space. To fill this gap, we first propose a rule that the deep embeddings of rotated images should be closer to each other than those of any other images (including the images belonging to the same class). Then, we propose to maximize the joint probability of the leave-one-out image classification and rotational image identification. With the assumption of independence, such optimization leads to the minimization of a novel loss function composed of two terms: 1) a class-discrimination term and 2) a rotation-invariant term. Furthermore, we introduce a penalty parameter that balances these two terms and further propose a final loss to Rotation-invariant Deep embedding for RS images, termed RiDe. Extensive experiments conducted on two benchmark RS datasets validate the effectiveness of the proposed approach and demonstrate its superior performance when compared to other state-of-the-art methods. The codes of this article will be publicly available at https://github.com/jiankang1991/TGRS_RiDe .

In the application of system identification, not only the output but also the input of the system may be corrupted by noise, which is often characterized by the errors-in-variables (EIV) model. To identify such systems, a gradient-descent total least-squares (GD-TLS) and a maximum total correntropy (MTC) algorithms were proposed. In some scenarios, the weight vector of the unknown system may be sparse, e.g., the echo path in acoustic echo cancelation (AEC). Employing TLS or MTC to estimate such systems may result in slow convergence rate, since they assign the same gain to the update of each weight and therefore cannot make use of the sparsity feature of the system to accelerate convergence. To address the above problem, this article proposes a uniform optimization model for deriving proportionate total adaptive filtering algorithms, and then two proportionate total adaptive filtering algorithms are developed, namely, the proportionate total normalized least mean square (PTNLMS) algorithm for Gaussian noise disturbance and the proportionate MTC (PMTC) algorithm for impulsive noise interference, which are both derived by utilizing the method of Lagrange multipliers. Moreover, this article also makes a steady-state performance analysis of the two proposed algorithms. Simulations are performed to demonstrate the superior performance of the two proposed algorithms and to test the accuracy of the theory on the steady-state performance analysis.