Noisy Target Research Articles

Unsupervised Domain Adaptation via Risk-Consistent Estimators

Unsupervised domain adaptation (UDA) attempts to learn domain invariant representations and has achieved significant progress, whereas self-training-based UDA methods have shown powerful performance. However, due to the domain gap, pseudo-labels selected through high confidence scores or uncertainty inevitably contain noise, leading to inaccurate predictions. To address this issue, we propose a novel risk-consistent training method. Specifically, both clean and noisy classifiers are introduced to estimate the noise transition matrix. The clean classifier is exploited to assign pseudo-labels for target data in each iteration. The noisy classifier is then trained with noisy target samples, and the optimal parameters are obtained through a closed-form solution. Heuristically, we also pre-train a domain predictor to select a target-like source example for the noise transition matrix estimation. In addition, we design an uncertainty-guided regularization to generate soft pseudo-labels and avoid overconfident predictions. Extensive experimental results show the effectiveness of our method, and state-of-the-art performance has been achieved. Codes are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/feifei-cv/RCE.</uri>

Frequency-Domain Deep Guided Image Denoising

Despite the tremendous advances in image denoising techniques, there still remains a challenge to restore a clean image with salient structures based only on one single noisy observation, especially at high noise levels. In this work, to address this issue, we propose a novel frequency-domain guided denoising algorithm to conduct denoising with the help of an additional well-aligned guidance image. Thanks to the structural correlation between the input paired images, the frequency characteristics of the guidance image can indicate whether the frequency coefficients of the noisy target image are contributed by noise or textures. Therefore, the explicit frequency decomposition enables our model to avoid over-smoothing those detailed contents when removing noise. However, as two input images are usually captured in different fields, their structures are not always consistent with each other. Therefore, we model guided denoising with an optimization problem which takes into account both the representation model with respect to the guidance image and the structural fidelity to the noisy target image. Based on this, we further design a convolutional neural network, called as FGDNet, to explore the optimal solution. Due to the visual masking phenomenon, the human visual system is sensitive to noise in the flat areas, but may not perceive noise around edges or textures. Therefore, to handle a denoising task, we expect to remove as much noise as possible to guarantee the spatial smoothness of the flat contents, while preserving the high-frequency structures as well. Through frequency decomposition, our model can process the low-frequency and high-frequency contents separately with different denoising schemes. In the training stage, we also adopt a frequency-relevant loss function to optimize the network. Experimental results show that, compared with the state-of-the-art guided and non-guided denoising algorithms, our FGDNet can achieve higher denoising accuracy and also better visual quality in both flat and texture-rich regions. Source codes are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/lustrouselixir/FGDNet</uri> .

A model-driven network for guided image denoising

Guided image denoising recovers clean target images by fusing guidance images and noisy target images. Several deep neural networks have been designed for this task, but they are black-box methods lacking interpretability. To overcome the issue, this paper builds a more interpretable network. To start with, an observation model is proposed to account for modality gap between target and guidance images. Then, this paper formulates a deep prior regularized optimization problem, and solves it by alternating direction method of multipliers (ADMM) algorithm. The update rules are generalized to design the network architecture. Extensive experiments conducted on FAIP and RNS datasets manifest that the novel network outperforms several state-of-the-art and benchmark methods regarding both evaluation metrics and visual inspection.

Fully Unsupervised Machine Translation Using Context-Aware Word Translation and Denoising Autoencoder

ABSTRACT Learning machine translation by using only monolingual data sets is a complex task as there are many possible ways to connect or associate target sentences with source sentences. The monolingual word embeddings are linearly mapped on a common shared space through robust learning or adversarial training in an unsupervised way, but these learning techniques have fundamental limitations in translating sentences. In this paper, a simple yet effective method has been proposed for fully unsupervised machine translation that is based on cross-lingual sense to word embedding instead of cross-lingual word embedding and language model. We have utilized word sense disambiguation to incorporate the source language context in order to select the sense of a word more appropriately. A language model for considering target language context in lexical choices and denoising autoencoder for language insertion, deletion, and reordering are integrated. The proposed approach eliminates the problem of noisy target language context due to erroneous word translations. This work takes into account the challenge of homonyms and polysemous words in the case of morphologically rich languages. The experiments performed on English-Hindi and Hindi-English using different evaluation metrics show an improvement of +3 points in BLEU and METEOR-Hindi over the baseline system.

Unsupervised CNN Based on Self-Similarity for Seismic Data Denoising

Convolutional neural network (CNN)-based methods are powerful tools for seismic data denoising. Most methods adopt a supervised learning strategy, which requires noise-free labels to construct an objective function to guide the training of network parameters; however, it is impossible to obtain true noise-free field data. We propose a novel unsupervised random-noise-suppression method that can train a network directly on noisy target data without noise-free labels. The proposed method is inspired by the simple denoising idea of averaging multiple noisy observations, and it requires noise to satisfy two assumptions: it should be zero-mean and independent of the signal. In this method, multiple observations can be used as labels, and the loss function is constructed as the mean square error expected between the network output and these observations. The trained network estimates the expected value of these noise observations (i.e., the clean signal). This unsupervised method theoretically requires multiple noisy observations. Considering the good nonlocal self-similarity of seismic data, we used self-similar blocks to rearrange the data to construct multiple pseudo-observations and finally realize unsupervised training. The proposed method was compared with the traditional <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">f-x</i> deconvolution, curvelet, and generator CNN method on synthetic and field data, and the experimental results verified the effectiveness of the proposed method.

Joint image denoising with gradient direction and edge-preserving regularization

Joint image denoising algorithms use the structures of the guidance image as a prior to restore the noisy target image. While the provided guidance images are helpful to improve the denoising performance, the denoised edges are most likely to be blurred especially when the edges of the guidance image are weak or inexistent. To address this weakness, this paper proposes a new gradient-direction-based joint image denoising method in which the absolute cosine value of the angle between two gradient vectors of the guidance image and those of the image to recover is employed as the parallel measurement to ensure that the gradient directions of the denoised image are approximately the same as or opposite to those of the guidance image. Besides, a new edge-preserving regularization term is developed to alleviate the effects of the unreliable prior information from guidance image. To simplify the resultant complex nonconvex and nonlinear fractional model, the logarithm function is employed to convert the multiplication operation into addition operation. Then, we construct the surrogate function for the logarithmic term of l2-norm, and separate the variables to transform the objective function into convex one with high numerical stability while retaining high efficiency. Finally, the optimal solutions can be obtained by directly minimizing the convex functions. Experimental results on public datasets and from nine benchmark methods consistently demonstrate the effectiveness of the proposed method both visually and quantitatively.

Prior context influences motor brain areas in an auditory oddball task and prefrontal cortex multitasking modelling

In this study, the relationship of orienting of attention, motor control and the Stimulus- (SDN) and Goal-Driven Networks (GDN) was explored through an innovative method for fMRI analysis considering all voxels in four experimental conditions: standard target (Goal; G), novel (N), neutral (Z) and noisy target (NG). First, average reaction times (RTs) for each condition were calculated. In the second-level analysis, ‘distracted’ participants, as indicated by slower RTs, evoked brain activations and differences in both hemispheres’ neural networks for selective attention, while the participants, as a whole, demonstrated mainly left cortical and subcortical activations. A context analysis was run in the behaviourally distracted participant group contrasting the trials immediately prior to the G trials, namely one of the Z, N or NG conditions, i.e. Z.G, N.G, NG.G. Results showed different prefrontal activations dependent on prior context in the auditory modality, recruiting between 1 to 10 prefrontal areas. The higher the motor response and influence of the previous novel stimulus, the more prefrontal areas were engaged, which extends the findings of hierarchical studies of prefrontal control of attention and better explains how auditory processing interferes with movement. Also, the current study addressed how subcortical loops and models of previous motor response affected the signal processing of the novel stimulus, when this was presented laterally or simultaneously with the target. This multitasking model could enhance our understanding on how an auditory stimulus is affecting motor responses in a way that is self-induced, by taking into account prior context, as demonstrated in the standard condition and as supported by Pulvinar activations complementing visual findings. Moreover, current BCI works address some multimodal stimulus-driven systems.

Multilevel Wavelet-SRNet for SAR Target Recognition

Speckle noise is an important factor affecting the accuracy of synthetic aperture radar (SAR) target recognition. Traditional speckle reduction methods based on transform domain and spatial filtering usually require professional experience to set the threshold, which will also affect the recognition accuracy. This letter proposes a multilevel wavelet speckle reduction network (Wavelet-SRNet) for noisy SAR images target recognition. First, the method designs the wavelet soft threshold denoising method as a trainable neural network module in the convolutional neural network (CNN) framework. Then, a two-level wavelet denoising branch is constructed and fused with the original noisy image. Finally, we cascade a CNN-based classification model on the above structure to form an SAR image target recognition network whose denoising threshold can be automatically learned. Experiments on the moving and stationary target acquisition and recognition database show that the classification accuracy of the proposed method for target recognition in noisy SAR images is better than the compared state-of-the-art methods. Also, the method achieved high test accuracy in the noise augmentation experiment.

Quantum illumination via quantum-enhanced sensing

Quantum-enhanced sensing has a goal of enhancing a parameter sensitivity with input quantum states, while quantum illumination has a goal of enhancing a target detection capability with input entangled states in a heavy noise environment. Here we propose a concatenation between quantum-enhanced sensing and quantum illumination that can take quantum advantage over the classical limit. First, phase sensing in an interferometry is connected to a target sensing via quantum Fisher information. Second, the target sensitivity is investigated in noisy quantum-enhanced sensing. Under the same input state energy, for example, $N$-photon entangled states can exhibit better performance than a two-mode squeezed vacuum state and a separable coherent state. Incorporating a photon-number difference measurement, finally, the noisy target sensitivity is connected to a signal-to-noise ratio which is associated with a minimum error probability of discriminating the presence and absence of the target. We show that both the target sensitivity and the signal-to-noise ratio can be enhanced with increasing thermal noise.

Evaluating recall error in preschoolers: Category expectations influence episodic memory for color

Despite limited memory capacity, children are exceptional learners. How might children engage in meaningful learning despite limited memory systems? Past research suggests that adults integrate category knowledge and noisy episodic traces to aid recall when episodic memory is noisy or incomplete (e.g. Hemmer & Steyvers, 2009a,b). We suspect children utilize a similar process but integrate category and episodic traces in recall to a different degree. Here we conduct two experiments to empirically assess children’s color category knowledge (Study 1) and recall of target hue values (Study 2). In Study 1, although children’s generated hue values appear to be noisier than adults, we found no significant difference between children and adult’s generated color category means (prototypes), suggesting that preschool-aged children’s color categories are well established. In Study 2, we found that children’s (like adult’s) free recall of target hue values regressed towards color category means. We implemented three probabilistic memory models: one that combines category knowledge and specific target information (Integrative), a category only (Noisy Prototype) model, and a target only (Noisy Target) model to computationally evaluate recall performance. Consistent with previous studies with older children (Duffy, Huttenlocher, & Crawford, 2006), quantitative fits of the models to aggregate group-level data provided strong support for the Integrative process. However, at the individual subject level, a greater proportion of preschoolers’ recall was better fit by a Prototype only model. Our results provide evidence that the integration of category knowledge in episodic memory comes online early and strongly. Implications for how the greater reliance on category knowledge by preschoolers relative to adults might track with developmental shifts in relational episodic memory are discussed.

The role of priors in controlling smooth pursuit eye movements of clear and noisy target motions: predictive pursuit does more than compensate for time delays

The role of priors in controlling smooth pursuit eye movements of clear and noisy target motions: predictive pursuit does more than compensate for time delays

Influence of prior and visual information on eye movements in amblyopic children.

This study analyzed the characteristics of pursuit and assessed the influence of prior and visual information on eye velocity and saccades in amblyopic and control children, in comparison to adults. Eye movements of 41 children (21 amblyopes and 20 controls) were compared to eye movements of 55 adults (18 amblyopes and 37 controls). Participants were asked to pursue a target moving at a constant velocity. The target was either a 'standard' target, with a uniform color intensity, or a 'noisy' target, with blurry edges, to mimic the blurriness of an amblyopic eye. Analysis of pursuit patterns showed that the onset was delayed, and the gain was decreased in control children with a noisy target in comparison to amblyopic or control children with a standard target. Furthermore, a significant effect of prior and visual information on pursuit velocity and saccades was found across all participants. Moreover, the modulation of the effect of visual information on the pursuit velocity by group, that is amblyopes or controls with a standard target, and controls with a noisy target, was more limited in children. In other words, the effect of visual information was higher in control adults with a standard target compared to control children with the same target. However, in the case of a blurry target, either in control participants with a noisy target or in amblyopic participants with a standard target, the effect of visual information was larger in children.

Coarse-Grained Diffraction Template Matching Model to Retrieve Multiconformational Models for Biomolecule Structures from Noisy Diffraction Patterns.

Biomolecular imaging using X-ray free-electron lasers (XFELs) has been successfully applied to serial femtosecond crystallography. However, the application of single-particle analysis for structure determination using XFELs with 100 nm or smaller biomolecules has two practical problems: the incomplete diffraction data sets for reconstructing 3D assembled structures and the heterogeneous conformational states of samples. A new diffraction template matching method is thus presented here to retrieve a plausible 3D structural model based on single noisy target diffraction patterns, assuming candidate structures. Two concepts are introduced here: prompt candidate diffraction, generated by enhanced sampled coarse-grain (CG) candidate structures, and efficient molecular orientation searching for matching based on Bayesian optimization. A CG model-based diffraction-matching protocol is proposed that achieves a 100-fold speed increase compared to exhaustive diffraction matching using an all-atom model. The conditions that enable multiconformational analysis were also investigated by simulated diffraction data for various conformational states of chromatin and ribosomes. The proposed method can enable multiconformational analysis, with a structural resolution of at least 20 Å for 270-800 Å flexible biomolecules, in experimental single-particle structure analyses that employ XFELs.

Low-Dose CT Image Restoration Based on Adaptive Prior Feature Matching and Nonlocal Means

Although the low-dose CT (LDCT) technique can reduce the radiation damage to patients, it will be highly detrimental to the reconstructed image quality. The normal-dose scan assisted algorithms have shown their potential in improving LDCT image quality by using a registered previously scanned normal-dose CT (NDCT) reference to regularize the corresponding LDCT target. The major drawback of such methods is the requirement of a previous patient-specific NDCT scan, which limits their clinical application. To address these problems, this paper proposed adaptive prior feature matching method for better restoration of the LDCT image. The innovation lies in construction of offline texture feature database and online adaptive prior feature matching integrated with the NLM regularization. Specifically, the prior features were extracted by the gray level co-occurrence matrix (GLCM) from regions of interest (ROIs) in existing NDCT scans of population patients. For online adaptive prior feature matching, ROIs with their texture features being similar to those of the current noisy target ROI are selected from the database as the references for the NLM regularization. The effectiveness of the proposed algorithm is validated by clinical lung cancer studies, the gain over traditional methods is noticeable in terms of both noise suppression and textures preservation.

Study a bearing-only moving ground target tracking problem using single seismic sensor

The problem of tracking moving ground targets using seismic sensors is considered in this paper. Noisy seismic data induced from a moving ground vehicle is detected and collected by a single, fixed, and passive three-component seismic sensor. Two Bayesian suboptimal estimator, namely the Extended Kalman filter (EKF) and the Unscented Kalman filter (UKF), and the optimal Monte Carlo based particle filter (PF) were used in estimating and tracking the true angular behavior of the target. The comparison between these estimators showed that they have almost the same accuracy in estimating the mean value of the noisy target azimuth. In terms of filter consistency, EKF and PF with a number of particles (NP = 5000) are superior to the UKF estimator.

Altitude Estimation Using Multipath With a Two-Dimensional Radar Over Spherical Earth

This paper presents a target altitude estimation approach based on multipath detections with range and range rate information from a two-dimensional (2-D) airborne radar. Using multipath detections has to deal with the following challenges: first, low probability of detection (PD) of the indirect path signal reflected by the sea surface; second, lack of an analytical expression for the relationship between the target motion parameter/state and the indirect path measurements over spherical earth; and third, unknown range measurement error standard deviation (s.d.) for the indirect path signal. Two estimators are developed: an adaptive multiple model iterated least squares (AMM-ILS) estimator for motion parameter estimation (noiseless target motion) and an AMM unscented Kalman filter (AMM-UKF) for dynamic state estimation (noisy target motion). Altitude estimate accuracy is studied based on simulation data. It shows that the AMM-ILS yields statistically efficient estimates (i.e., optimal and with a quantifiable accuracy), but this is not always true for the AMM-UKF. For a long distant target (e.g., 300 km) with very low indirect path PD (0.1 and 0.2), the AMM-UKF is inconsistent and inefficient statistically, due to biased estimation. The altitude Cramer Rao lower bounds from a 3-D radar with range and elevation measurements are also computed for comparison. The results show that the 2-D radar can provide better altitude estimates than the 3-D radar in most of the cases, except when the 3-D radar has accurate elevation measurement (with error s.d. $\sigma _e=0.5^{\circ }$ ) and the 2-D radar has very low indirect path PD (PD = 0.1).

Low-Dose Lung CT Image Restoration Using Adaptive Prior Features From Full-Dose Training Database

The valuable structure features in full-dose computed tomography (FdCT) scans can be exploited as prior knowledge for low-dose CT (LdCT) imaging. However, lacking the capability to represent local characteristics of interested structures of the LdCT image adaptively may result in poor preservation of details/textures in LdCT image. This paper aims to explore a novel prior knowledge retrieval and representation paradigm, called adaptive prior features assisted restoration algorithm, for the purpose of better restoration of the low-dose lung CT images by capturing local features from FdCT scans adaptively. The innovation lies in the construction of an offline training database and the online patch-search scheme integrated with the principal components analysis (PCA). Specifically, the offline training database is composed of 3-D patch samples extracted from existing full-dose lung scans. For online patch-search, 3-D patches with structure similar to the noisy target patch are first selected from the database as the training samples. Then, PCA is applied on the training samples to retrieve their local prior principal features adaptively. By employing the principal features to decompose the noisy target patch and using an adaptive coefficient shrinkage technique for inverse transformation, the noise of the target patch can be efficiently removed and the detailed texture can be well preserved. The effectiveness of the proposed algorithm was validated by CT scans of patients with lung cancer. The results show that it can achieve a noticeable gain over some state-of-the-art methods in terms of noise suppression and details/textures preservation.

Dynamic causal modelling of eye movements during pursuit: Confirming precision-encoding in V1 using MEG

This paper shows that it is possible to estimate the subjective precision (inverse variance) of Bayesian beliefs during oculomotor pursuit. Subjects viewed a sinusoidal target, with or without random fluctuations in its motion. Eye trajectories and magnetoencephalographic (MEG) data were recorded concurrently. The target was periodically occluded, such that its reappearance caused a visual evoked response field (ERF). Dynamic causal modelling (DCM) was used to fit models of eye trajectories and the ERFs. The DCM for pursuit was based on predictive coding and active inference, and predicts subjects' eye movements based on their (subjective) Bayesian beliefs about target (and eye) motion. The precisions of these hierarchical beliefs can be inferred from behavioural (pursuit) data. The DCM for MEG data used an established biophysical model of neuronal activity that includes parameters for the gain of superficial pyramidal cells, which is thought to encode precision at the neuronal level. Previous studies (using DCM of pursuit data) suggest that noisy target motion increases subjective precision at the sensory level: i.e., subjects attend more to the target's sensory attributes. We compared (noisy motion-induced) changes in the synaptic gain based on the modelling of MEG data to changes in subjective precision estimated using the pursuit data. We demonstrate that imprecise target motion increases the gain of superficial pyramidal cells in V1 (across subjects). Furthermore, increases in sensory precision – inferred by our behavioural DCM – correlate with the increase in gain in V1, across subjects. This is a step towards a fully integrated model of brain computations, cortical responses and behaviour that may provide a useful clinical tool in conditions like schizophrenia.

Multi-view radar target recognition based on multitask compressive sensing

A novel multitask compressive sensing (MtCS)-based method for multi-view radar automatic target recognition is presented in the paper. The sparse representation vectors recovered jointly via MtCS are used as recognition features, and classification is performed according to minimum reconstruction error criterion. Compared to the conventional methods, the proposed method has a significant advantage of exploiting the statistical correlation among multiple views for target recognition. Experiments were conducted using a synthetic vehicle target data-set and the moving and stationary target acquisition and recognition database. The results show that the proposed method achieves promising recognition accuracy, and is robust with respect to noisy observations and complex target types.

Word segmentation and pronunciation extraction from phoneme sequences through cross-lingual word-to-phoneme alignment

In this paper, we study methods to discover words and extract their pronunciations from audio data for non-written and under-resourced languages. We examine the potential and the challenges of pronunciation extraction from phoneme sequences through cross-lingual word-to-phoneme alignment. In our scenario a human translator produces utterances in the (non-written) target language from prompts in a resource-rich source language. We add the resource-rich source language prompts to help the word discovery and pronunciation extraction process. By aligning the source language words to the target language phonemes, we segment the phoneme sequences into word-like chunks. The resulting chunks are interpreted as putative word pronunciations but are very prone to alignment and phoneme recognition errors. Thus we suggest our alignment model Model 3P that is particularly designed for cross-lingual word-to-phoneme alignment. We present two different methods (source word dependent and independent clustering) that extract word pronunciations from word-to-phoneme alignments and compare them. We show that both methods compensate for phoneme recognition and alignment errors. We also extract a parallel corpus consisting of 15 different translations in 10 languages from the Christian Bible to evaluate our alignment model and error recovery methods. For example, based on noisy target language phoneme sequences with 45.1% errors, we build a dictionary for an English Bible with a Spanish Bible translation with 4.5% OOV rate, where 64% of the extracted pronunciations contain no more than one wrong phoneme. Finally, we use the extracted pronunciations in an automatic speech recognition system for the target language and report promising word error rates – given that pronunciation dictionary and language model are learned completely unsupervised and no written form for the target language is required for our approach.