Search

Noisy labels often occur in vision datasets, especially when they are obtained from crowdsourcing or Web scraping. We propose a new regularization method, which enables learning robust classifiers in presence of noisy data. To achieve this goal, we propose a new adversarial regularization scheme based on the Wasserstein distance. Using this distance allows taking into account specific relations between classes by leveraging the geometric properties of the labels space. Our Wasserstein Adversarial Regularization (WAR) encodes a selective regularization, which promotes smoothness of the classifier between some classes, while preserving sufficient complexity of the decision boundary between others. We first discuss how and why adversarial regularization can be used in the context of noise and then show the effectiveness of our method on five datasets corrupted with noisy labels: in both benchmarks and real datasets, WAR outperforms the state-of-the-art competitors. [ABSTRACT FROM AUTHOR]

Space shift keying (SSK) is a single radio frequency (RF) chain multiple-input multiple-output (MIMO) scheme that uses the index of transmit antennas for information transmission. Time successive SSK- $ M$ -ary modulation (TSSM) and multi RF chain TSSM (MRF-TSSM) are recent MIMO techniques that achieve second order transmit diversity. We propose two cooperative relaying systems to attain distributed diversity in addition to transmit diversity of TSSM and MRF-TSSM. Cooperative-TSSM-1 (Coop-TSSM-1) uses multibranch relaying where TSSM is used in orthogonal time slots during the cooperative phase. Coop-TSSM-2 maps symbols to lower order constellations at multiple relays for simultaneous transmission of information using MRF-TSSM. We consider Decode-and-forward (DF) relaying where relays that correctly detect symbols forward it. We analyze the average number of relays that participate in cooperation, the end-to-end performance using a tight upper bound on the bit error rate (BER) and the asymptotic BER expressions to analyze diversity orders of both cooperative systems. We validate all analyzes using simulation results. From the analysis and results we show that Coop-TSSM-1 achieves better diversity order compared to Coop-TSSM-2 at the cost of higher number of time slots for completing the cooperative phase. Further in Coop-TSSM-2, results indicate that different constellation mappings used in MRF-TSSM during cooperative phase impact the BER performance. By applying spatial modulation based transmit diversity and $M$ -ary Modulation systems in the cooperative setups of Coop-TSSM schemes, performance comparison is given which highlights the significance of Coop-TSSM schemes. [ABSTRACT FROM AUTHOR]

A two-timeslot N\text{RF} radio frequency (RF) chain transmitter is proposed in this work that takes advantage of the independence of sum and difference of Gaussian random vectors to achieve second-order transmit diversity. Each RF chain in the transmitter is equipped with $N_t$ transmitter antennas such that the transmitter is capable of switching the modulation between space shift keying (SSK) and either M$-ary phase shift keying (MPSK) or M$-ary quadrature amplitude modulation (MQAM). In timeslot 1, the proposed multi-RF chain time successive SSK-M information symbols. In timeslot 2, the same block of information symbols are transmitted using either MPSK or MQAM by activating all $ N_t$ antennas of every RF chain. For this system, the maximal likelihood detection metric, transmit diversity order, coding gain, and computational complexity are derived. The performance of MRF-TSSM is analyzed considering transmitter correlation using a tight upper bound of the bit error rate. The analysis is validated using simulation results, which show that MRF-TSSM achieves significant performance gains compared to some existing transmit diversity systems, and the transmit diversity of MRF-TSSM degrades only slightly even for high transmitter correlation. Furthermore, the performance of MRF-TSSM is studied with imperfect channel state information at the receiver. [ABSTRACT FROM AUTHOR]

Designing an effective criterion to select a subset of features is a challenging problem for hyperspectral image classification. In this paper, we develop a feature selection method to select a subset of class discriminant features for hyperspectral image classification. First, we propose a new class separability measure based on the surrogate kernel and Hilbert Schmidt independence criterion in the reproducing kernel Hilbert space. Second, we employ the proposed class separability measure as an objective function and we model the feature selection problem as a continuous optimization problem using LASSO optimization framework. The combination of the class separability measure and the LASSO model allows selecting the subset of features that increases the class separability information and also avoids a computationally intensive subset search strategy. Experiments conducted with three hyperspectral data sets and different experimental settings show that our proposed method increases the classification accuracy and outperforms the state-of-the-art methods. [ABSTRACT FROM PUBLISHER]

Wireless microsensors in hospitals, factories, and farms can manage and save lives and resources. Although tiny batteries cannot sustain them for long, harvesters can because ambient energy is abundant. Photovoltaic cells are popular in this regard because they output close to 100× more power from solar light than piezoelectric, electrostatic, and thermoelectric generators can from motion and heat. However, since mm cells can only supply μW's of the mW's that microsystems can draw, and light is not always available, battery assistance is necessary. So to supply functions and sustain operation across extended periods, the system should extract maximum ambient power, draw minimal battery assistance, and deliver as much power as possible. The 0.18-μm CMOS harvester presented here does this, draws 10–100 μW from a 3 \times 3 \times 1 mm^3 cell and assistance from a battery to supply a 1-mW load and recharge the battery with excess cell power. The switched-inductor charger supply regulates 1 V within $\pm 25$ mV with 73\-86\% power-conversion efficiency and keeps the cell within 1% of its maximum power point. This way, the cell outputs 100\, {\rm \mu W/mm}^{2} from solar light and 1 μW/mm2 from direct indoor light. [ABSTRACT FROM PUBLISHER]

Partial shading is one of the certain conditions in photovoltaic (PV) power plants. Often the panels get partially shaded due to soiling, clouds, and trees. The effect of shading is of high concern and importance especially in applications such as water pumps due to agricultural environments in which they are employed in. However, the research focus on PV fed pumps till date has been restricted to only uniform shading conditions. Unlike uniform shading conditions during maximum power tracking, where the panel voltage remains almost constant for an entire range of irradiance, partial shading offer conditions such as highly variable panel voltage and transitions of intermediate DC–DC power converter from continuous conduction mode to discontinuous conduction mode. These effects severely affect the performance of the power converter and, therefore, the power output of the pump. This paper presents a study on the effects caused by partial shading conditions on pumps through simulations and verified by experimentations. The simulation and experimental results are found to be in good agreement with each other. This research thus helps in understanding the detrimental effects caused by partial shading conditions and thus serves as a reference tool for practitioners who wish to study PV fed pumps. [ABSTRACT FROM AUTHOR]

Characteristics of photovoltaic arrays exhibit multiple peaks under mismatching conditions. In order to harness maximum energy, it is imperative to track the global maximum power point. A novel global peak tracking algorithm is proposed using current control in this paper. The proposed method operates in the backward phase and forward phase to track the global peak. The objective of the proposed algorithm is to track the global maximum accurately with minimum tracking time. The enhanced performance of the proposed algorithm is verified using simulations by comparing it with an existing method. Experimental validation is done using a solar array simulator, boost converter, resistive load, and dSPACE controller. Experimental results are in close agreement with simulation. The proposed algorithm is intended to track the global peak of a PV string that contains a group of modules. [ABSTRACT FROM AUTHOR]