Deep Transfer Learning for Cerebral Cortex Using Area-Preserving Geometry Mapping.

Limited sample size hinders the application of deep learning in brain image analysis, and transfer learning is a possible solution. However, most pretrained models are 2D based and cannot be applied directly to 3D brain images. In this study, we propose a novel framework to apply 2D pretrained models to 3D brain images by projecting surface-based cortical morphometry into planar images using computational geometry mapping. Firstly, 3D cortical meshes are reconstructed from magnetic resonance imaging (MRI) using FreeSurfer and projected into 2D planar meshes with topological preservation based on area-preserving geometry mapping. Then, 2D deep models pretrained on ImageNet are adopted and fine-tuned for cortical image classification on morphometric shape metrics. We apply the framework to sex classification on the Human Connectome Project dataset and autism spectrum disorder (ASD) classification on the Autism Brain Imaging Data Exchange dataset. Moreover, a 2-stage transfer learning strategy is suggested to boost the ASD classification performance by using the sex classification as an intermediate task. Our framework brings significant improvement in sex classification and ASD classification with transfer learning. In summary, the proposed framework builds a bridge between 3D cortical data and 2D models, making 2D pretrained models available for brain image analysis in cognitive and psychiatric neuroscience.

Joint Embedding Learning and Low-Rank Approximation: A Framework for Incomplete Multiview Learning.

In real-world applications, not all instances in the multiview data are fully represented. To deal with incomplete data, incomplete multiview learning (IML) rises. In this article, we propose the joint embedding learning and low-rank approximation (JELLA) framework for IML. The JELLA framework approximates the incomplete data by a set of low-rank matrices and learns a full and common embedding by linear transformation. Several existing IML methods can be unified as special cases of the framework. More interestingly, some linear transformation-based complete multiview methods can be adapted to IML directly with the guidance of the framework. Thus, the JELLA framework improves the efficiency of processing incomplete multiview data, and bridges the gap between complete multiview learning and IML. Moreover, the JELLA framework can provide guidance for developing new algorithms. For illustration, within the framework, we propose the IML with the block-diagonal representation (IML-BDR) method. Assuming that the sampled examples have an approximate linear subspace structure, IML-BDR uses the block-diagonal structure prior to learning the full embedding, which would lead to more correct clustering. A convergent alternating iterative algorithm with the successive over-relaxation optimization technique is devised for optimization. The experimental results on various datasets demonstrate the effectiveness of IML-BDR.

Multiview Classification With Cohesion and Diversity.

Different views of multiview data share certain common information (consensus) and also contain some complementary information (complementarity). Both consensus and complementarity are of significant importance to the success of multiview learning. In this paper, we explicitly formulate both of them for multiview classification. On the one hand, a cohesion-increasing loss term with a learnable label-adjusting matrix is designed to facilitate consensus among views in the training stage. With this kind of loss, the learned classifiers of all views are more likely to obtain the correct classification, thereby maximizing the agreement among views. On the other hand, an independence measurement is adopted as the diversity-promoting regularization to encourage classifiers to be diverse such that more complementary information can be captured by these "diversified" classifiers. Overall, the resultant model is capable of achieving more comprehensive and accurate classification by exploring and exploiting the common and complementary information across multiple views more thoroughly. An iterative optimization algorithm with proved convergence is proposed for training the model. Extensive experimental results on various datasets have demonstrated the efficacy of the proposed method.

Single-shot coherent power-spectrum imaging of objects hidden by opaque scattering media.

We report coherent imaging of objects behind opaque scattering media with only one piece of the power spectrum pattern. We solve the unique solution and improve algorithm speed for the inverse problem. Based on the proposed scattering-disturbance model, with only one piece of the Fourier transform power spectrum pattern under coherent illumination, we successfully reconstruct clear images of the objects fully hidden by an opaque diffuser. The experimental results demonstrate the feasibility of the reconstruction method and the scattering-disturbance model. Our method makes it possible to carry out snapshot coherent imaging of the objects obscured by scattering media, which extends the methodology of x-ray crystallography to visible-light scattering imaging for underwater and living biomedical imaging.

Fast Detection of a Weak Signal by a Stochastic Resonance Induced by a Coherence Resonance in an Excitable GaAs/Al_{0.45}Ga_{0.55}As Superlattice.

The effect of a coherence resonance is observed experimentally in a GaAs/Al_{0.45}Ga_{0.55}As superlattice under dc bias at room temperature, which is driven by noise. For an applied voltage, for which no current self-oscillations are observed, regular current self-oscillations with a frequency of about 82 MHz are induced by exceeding a certain noise amplitude. In addition, a novel kind of a stochastic resonance is identified, which is triggered by the coherence resonance. This stochastic resonance appears when the device is driven by an external ac signal with a frequency, which is relatively close to that of the regular current self-oscillations at the coherence resonance. The intrinsic oscillation mode in the coherence resonance is found to be phase locked by an extremely weak ac signal. It is demonstrated that an excitable superlattice device can be used for the fast detection of weak signals submerged in noise. These results are very well reproduced by results using numerical simulations based on a sequential resonant tunneling model of nonlinear electron transport in semiconductor superlattices.

Design of direct-vision cyclo-olefin-polymer double Amici prism for spectral imaging.

A direct-vision Amici prism is a desired dispersion element in the value of spectrometers and spectral imaging systems. In this paper, we focus on designing a direct-vision cyclo-olefin-polymer double Amici prism for spectral imaging systems. We illustrate a designed structure: E48R/N-SF4/E48R, from which we obtain 13 deg dispersion across the visible spectrum, which is equivalent to 700 line pairs/mm grating. We construct a simulative spectral imaging system with the designed direct-vision cyclo-olefin-polymer double Amici prism in optical design software and compare its imaging performance to a glass double Amici prism in the same system. The results of spot-size RMS demonstrate that the plastic prism can serve as well as their glass competitors and have better spectral resolution.

Restoration of longitudinal laser tomography target image from inhomogeneous medium degradation under common conditions.

In order to overcome the shortages of the target image restoration method for longitudinal laser tomography using self-calibration, a more general restoration method through backscattering medium images associated with prior parameters is developed for common conditions. The system parameters are extracted from pre-calibration, and the LIDAR ratio is estimated according to the medium types. Assisted by these prior parameters, the degradation caused by inhomogeneous turbid media can be established with the backscattering medium images, which can further be used for removal of the interferences of turbid media. The results of simulations and experiments demonstrate that the proposed image restoration method can effectively eliminate the inhomogeneous interferences of turbid media and achieve exactly the reflectivity distribution of targets behind inhomogeneous turbid media. Furthermore, the restoration method can work beyond the limitation of the previous method that only works well under the conditions of localized turbid attenuations and some types of targets with fairly uniform reflectivity distributions.

Scalable Multi-View Semi-Supervised Classification via Adaptive Regression.

With the advent of multi-view data, multi-view learning has become an important research direction in machine learning and image processing. Considering the difficulty of obtaining labeled data in many machine learning applications, we focus on the multi-view semi-supervised classification problem. In this paper, we propose an algorithm named multi-view semi-supervised classification via adaptive regression (MVAR) to address this problem. Specifically, regression-based loss functions with l2,1 matrix norm are adopted for each view and the final objective function is formulated as the linear weighted combination of all the loss functions. An efficient algorithm with proved convergence is developed to solve the non-smooth l2,1 -norm minimization problem. Regressing to class labels directly makes the proposed algorithm efficient in calculation and can be applied to large-scale data sets. The adaptively optimized weight coefficients balance the contributions of different views automatically, which makes the performance robust against the existence of low-quality views. With the learned projection matrices and bias vectors, predictions for out-of-sample data can be easily made. To validate the effectiveness of MVAR, comparisons are made with some benchmark methods on real-world data sets and in the scene classification scenario as well. The experimental results demonstrate the effectiveness of our proposed algorithm.