Search

Spatial resolution, spectral contrast and occlusion are three major bottlenecks for non-invasive inspection of complex samples with current imaging technologies. We exploit the sub-picosecond time resolution along with spectral resolution provided by terahertz time-domain spectroscopy to computationally extract occluding content from layers whose thicknesses are wavelength comparable. The method uses the statistics of the reflected terahertz electric field at subwavelength gaps to lock into each layer position and then uses a time-gated spectral kurtosis to tune to highest spectral contrast of the content on that specific layer. To demonstrate, occluding textual content was successfully extracted from a packed stack of paper pages down to nine pages without human supervision. The method provides over an order of magnitude enhancement in the signal contrast and can impact inspection of structural defects in wooden objects, plastic components, composites, drugs and especially cultural artefacts with subwavelength or wavelength comparable layers.

We study the behavior of algebraic connectivity in a weighted graph that is subject to site percolation, random deletion of the vertices. Using a refined concentration inequality for random matrices we show in our main theorem that the (augmented) Laplacian of the percolated graph concentrates around its expectation. This concentration bound then provides a lower bound on the algebraic connectivity of the percolated graph. As a special case for $(n,d,\lambda)$-graphs (i.e., $d$-regular graphs on $n$ vertices with non-trivial eigenvalues less than $\lambda$ in magnitude) our result shows that, with high probability, the graph remains connected under a homogeneous site percolation with survival probability $p\ge 1-C_{1}n^{-C_{2}/d}$ with $C_{1}$ and $C_{2}$ depending only on $\lambda/d$.

We introduce and analyze a new technique for model reduction for deep neural networks. While large networks are theoretically capable of learning arbitrarily complex models, overfitting and model redundancy negatively affects the prediction accuracy and model variance. Our Net-Trim algorithm prunes (sparsifies) a trained network layer-wise, removing connections at each layer by solving a convex optimization program. This program seeks a sparse set of weights at each layer that keeps the layer inputs and outputs consistent with the originally trained model. The algorithms and associated analysis are applicable to neural networks operating with the rectified linear unit (ReLU) as the nonlinear activation. We present both parallel and cascade versions of the algorithm. While the latter can achieve slightly simpler models with the same generalization performance, the former can be computed in a distributed manner. In both cases, Net-Trim significantly reduces the number of connections in the network, while also providing enough regularization to slightly reduce the generalization error. We also provide a mathematical analysis of the consistency between the initial network and the retrained model. To analyze the model sample complexity, we derive the general sufficient conditions for the recovery of a sparse transform matrix. For a single layer taking independent Gaussian random vectors of length $N$ as inputs, we show that if the network response can be described using a maximum number of $s$ non-zero weights per node, these weights can be learned from $\mathcal{O}(s\log N)$ samples.

We propose a flexible convex relaxation for the phase retrieval problem that operates in the natural domain of the signal. Therefore, we avoid the prohibitive computational cost associated with "lifting" and semidefinite programming (SDP) in methods such as PhaseLift and compete with recently developed non-convex techniques for phase retrieval. We relax the quadratic equations for phaseless measurements to inequality constraints each of which representing a symmetric "slab". Through a simple convex program, our proposed estimator finds an extreme point of the intersection of these slabs that is best aligned with a given anchor vector. We characterize geometric conditions that certify success of the proposed estimator. Furthermore, using classic results in statistical learning theory, we show that for random measurements the geometric certificates hold with high probability at an optimal sample complexity. Phase transition of our estimator is evaluated through simulations. Our numerical experiments also suggest that the proposed method can solve phase retrieval problems with coded diffraction measurements as well., Comment: Accepted in AISTATS 2017. Extended the discussion of related work and added a few more references. Clarified some of the statements and notations

This paper investigates conditions under which certain kinds of systems of bilinear equations have a unique structured solution. In particular, we look at when we can recover vectors $\boldsymbol{w},\boldsymbol{q}$ from observations of the form \[ y_{\ell} = <\boldsymbol{w},\boldsymbol{b}_{\ell}><\boldsymbol{c}_{\ell},\boldsymbol{q}>, \quad \ell = 1,\ldots,L, \] where $\boldsymbol{b}_\ell,\boldsymbol{c}_\ell$ are known. We show that if $\boldsymbol{w}\in\mathbb{C}^{M_1}$ and $\boldsymbol{q}\in\mathbb{C}^{M_2}$ are sparse, with no more than $K$ and $N$ nonzero entries, respectively, and the $\boldsymbol{b}_\ell,\boldsymbol{c}_\ell$ are generic, selected as independent Gaussian random vectors, then $\boldsymbol{w},\boldsymbol{q}$ are uniquely determined from \[ L \geq \mathrm{Const}\cdot (K+N)\log^5(M_1M_2) \] such equations with high probability. The key ingredient in our analysis is a uniform probabilistic bound on how far a random process of the form \[Z(\boldsymbol{X}) = \sum_{\ell=1}^L|\boldsymbol{b}_\ell^*\boldsymbol{X}\boldsymbol{c}_\ell|^2 \] deviates from its mean over a set of structured matrices $\boldsymbol{X}\in\mathcal{X}$. As both $\boldsymbol{b}_\ell$ and $\boldsymbol{c}_\ell$ are random, this is a specialized type of $4$th order chaos; we refer to $Z(\boldsymbol{X})$ as an {\em empirical chaos process}. Bounding this process yields a set of general conditions for when the map $\boldsymbol{X}\rightarrow \{\boldsymbol{b}_\ell^*\boldsymbol{X}\boldsymbol{c}_\ell\}_{\ell=1}^L$ is a restricted isometry over the set of matrices $\mathcal{X}$. The conditions are stated in terms of general geometric properties of the set $\mathcal{X}$, and are explicitly computed for the case where $\mathcal{X}$ is the set of matrices that are simultaneously sparse and low rank., Comment: A counter example suggests that the result in Theorem 2 may not be completely true. This requires a re-evaluation of the entire manuscript and the underlying analytical proofs. The authors have therefore decided to withdraw the paper at this point

In this paper we propose an energy-efficient camera-based gesture recognition system powered by light energy for "always on" applications. Low energy consumption is achieved by directly extracting gesture features from the compressed measurements, which are the block averages and the linear combinations of the image sensor's pixel values. The gestures are recognized using a nearest-neighbour (NN) classifier followed by Dynamic Time Warping (DTW). The system has been implemented on an Analog Devices Black Fin ULP vision processor and powered by PV cells whose output is regulated by TI's DC-DC buck converter with Maximum Power Point Tracking (MPPT). Measured data reveals that with only 400 compressed measurements (768x compression ratio) per frame, the system is able to recognize key wake-up gestures with greater than 80% accuracy and only 95mJ of energy per frame. Owing to its fully self-powered operation, the proposed system can find wide applications in "always-on" vision systems such as in surveillance, robotics and consumer electronics with touch-less operation.

Heavy sweep distortion induced by alignments and inter-reflections of layers of a sample is a major burden in recovering 2D and 3D information in time resolved spectral imaging. This problem cannot be addressed by conventional denoising and signal processing techniques as it heavily depends on the physics of the acquisition. Here we propose and implement an algorithmic framework based on low-rank matrix recovery and alternating minimization that exploits the forward model for THz acquisition. The method allows recovering the original signal in spite of the presence of temporal-spatial distortions. We address a blind-demodulation problem, where based on several observations of the sample texture modulated by an undesired sweep pattern, the two classes of signals are separated. The performance of the method is examined in both synthetic and experimental data, and the successful reconstructions are demonstrated. The proposed general scheme can be implemented to advance inspection and imaging applications in THz and other time-resolved sensing modalities.