A variety of real-life mobile sensing applications are becoming available, especially in the life-logging, fitness tracking and health monitoring domains. These applications use mobile sensors embedded in smart phones to recognize human activities in order to get a better understanding of human behavior. While progress has been made, human activity recognition remains a challenging task. This is partly due to the broad range of human activities as well as the rich variation in how a given activity can be performed. Using features that clearly separate between activities is crucial. In this paper, we propose an approach to automatically extract discriminative features for activity recognition. Specifically, we develop a method based on Convolutional Neural Networks (CNN), which can capture local dependency and scale invariance of a signal as it has been shown in speech recognition and image recognition domains. In addition, a modified weight sharing technique, called partial weight sharing, is proposed and applied to accelerometer signals to get further improvements. The experimental results on three public datasets, Skoda (assembly line activities), Opportunity (activities in kitchen), Actitracker (jogging, walking, etc.), indicate that our novel CNN-based approach is practical and achieves higher accuracy than existing state-of-the-art methods.

Inverse scattering problems of the reconstructions of physical properties of a medium from boundary measurements are substantially challenging ones. This work aims to verify the performance on experimentally collected data of a newly developed convexification method for a 3D coefficient inverse problem for the case of unknown objects buried in a sandbox. The measured backscatter data are generated by a point source moving along an interval of a straight line and the frequency is fixed. Using a special Fourier basis, the method of this work strongly relies on a new derivation of a boundary value problem for a system of coupled quasilinear elliptic equations. This problem, in turn, is solved via the minimization of a Tikhonov-like functional weighted by a Carleman weight function. Different from the continuous case, our weighted cost functional in the partial finite difference does not need the penalty term to gain the global convergence analysis. The numerical verification is performed using experimental data, which are raw backscatter data of the electric field. These data were collected using a microwave scattering facility at The University of North Carolina at Charlotte.

A 3-D inverse medium problem in the frequency domain is considered. Another name for this problem is Coefficient Inverse Problem. The goal is to reconstruct spatially distributed dielectric constants from scattering data. Potential applications are in detection and identification of explosive-like targets. A single incident plane wave and multiple frequencies are used. A new numerical method is proposed. A theorem is proved, which claims that a small neigborhood of the exact solution of that problem is reached by this method without any advanced knowledge of that neighborhood. We call this property of that numerical method "global convergence". Results of numerical experiments for the case of the backscattering data are presented.

The goal of this paper is to reconstruct spatially distributed dielectric constants from complexvalued scattered wave field by solving a 3D coefficient inverse problem for the Helmholtz equation at multi-frequencies. The data are generated by only a single direction of the incident plane wave. To solve this inverse problem, a globally convergent algorithm is analytically developed. We prove that this algorithm provides a good approximation for the exact coefficient without any a priori knowledge of any point in a small neighborhood of that coefficient. This is the main advantage of our method, compared with classical approaches using optimization schemes. Numerical results are presented for both computationally simulated data and experimental data. Potential applications of this problem are in detection and identification of explosive-like targets.

We analyze in this paper the performance of a newly developed globally convergent numerical method for a coefficient inverse problem for the case of multi-frequency experimental backscatter data associated to a single incident wave. These data were collected using a microwave scattering facility at the University of North Carolina at Charlotte. The challenges for the inverse problem under the consideration are not only from its high nonlinearity and severe ill-posedness but also from the facts that the amount of the measured data is minimal and that these raw data are contaminated by a significant amount of noise, due to a non-ideal experimental setup. This setup is motivated by our target application in detecting and identifying explosives. We show in this paper how the raw data can be preprocessed and successfully inverted using our inversion method. More precisely, we are able to reconstruct the dielectric constants and the locations of the scattering objects with a good accuracy, without using any advanced a priori knowledge of their physical and geometrical properties.