Semi-supervised SVMs for classification with unknown class proportions and a small labeled dataset

In the design of practical web page classification systems one often encounters a situation in which the labeled training set is created by choosing some examples from each class; but, the class proportions in this set are not the same as those in the test distribution to which the classifier will be actually applied. The problem is made worse when the amount of training data is also small. In this paper we explore and adapt binary SVM methods that make use of unlabeled data from the test distribution, viz., Transductive SVMs (TSVMs) and expectation regularization/constraint (ER/EC) methods to deal with this situation. We empirically show that when the labeled training data is small, TSVM designed using the class ratio tuned by minimizing the loss on the labeled set yields the best performance; its performance is good even when the deviation between the class ratios of the labeled training set and the test set is quite large. When the labeled training data is sufficiently large, an unsupervised Gaussian mixture model can be used to get a very good estimate of the class ratio in the test set; also, when this estimate is used, both TSVM and EC/ER give their best possible performance, with TSVM coming out superior. The ideas in the paper can be easily extended to multi-class SVMs and MaxEnt models.

Capacity of fading gaussian channel with an energy harvesting sensor node

Network life time maximization is becoming an important design goal in wireless sensor networks. Energy harvesting has recently become a preferred choice for achieving this goal as it provides near perpetual operation. We study such a sensor node with an energy harvesting source and compare various architectures by which the harvested energy is used. We find its Shannon capacity when it is transmitting its observations over a fading AWGN channel with perfect/no channel state information provided at the transmitter. We obtain an achievable rate when there are inefficiencies in energy storage and the capacity when energy is spent in activities other than transmission.

A constellation power allocation scheme for two-user Gaussian MAC

Constellation Constrained (CC) capacity regions of two-user Gaussian Multiple Access Channels (GMAC) have been recently reported, wherein introducing appropriate rotation between the constellations of the two users is shown to maximally enlarge the CC capacity region. Such a Non-Orthogonal Multiple Access (NO-MA) method of enlarging the CC capacity region is referred to as Constellation Rotation (CR) scheme. In this paper, we propose a novel NO-MA technique called Constellation Power Allocation (CPA) scheme to enlarge the CC capacity region of two-user GMAC. We show that the CPA scheme offers CC sum capacities equal (at low SNR values) or close (at high SNR values) to those offered by the CR scheme with reduced ML decoding complexity for some QAM constellations. For the CR scheme, code pairs approaching the CC sum capacity are known only for the class of PSK and PAM constellations but not for QAM constellations. In this paper, we design code pairs with the CPA scheme to approach the CC sum capacity for 16-QAM constellations. Further, the CPA scheme used for two-user GMAC with random phase offsets is shown to provide larger CC sum capacities at high SNR values compared to the CR scheme.

Speeded up Gaussian mixture model algorithm for background subtraction

Adaptive Gaussian Mixture Models (GMM) have been one of the most popular and successful approaches to perform foreground segmentation on multimodal background scenes. However, the good accuracy of the GMM algorithm comes at a high computational cost. An improved GMM technique was proposed by Zivkovic to reduce computational cost by minimizing the number of modes adaptively. In this paper, we propose a modification to his adaptive GMM algorithm that further reduces execution time by replacing expensive floating point computations with low cost integer operations. To maintain accuracy, we derive a heuristic that computes periodic floating point updates for the GMM weight parameter using the value of an integer counter. Experiments show speedups in the range of 1.33 - 1.44 on standard video datasets where a large fraction of pixels are multimodal.

On the generalized degrees of freedom of the K-user symmetric MIMO Gaussian interference channel

This work derives inner and outer bounds on the generalized degrees of freedom (GDOF) of the K-user symmetric MIMO Gaussian interference channel. For the inner bound, an achievable GDOF is derived by employing a combination of treating interference as noise, zero-forcing at the receivers, interference alignment (IA), and extending the Han-Kobayashi (HK) scheme to K users, depending on the number of antennas and the INR/SNR level. An outer bound on the GDOF is derived, using a combination of the notion of cooperation and providing side information to the receivers. Several interesting conclusions are drawn from the bounds. For example, in terms of the achievable GDOF in the weak interference regime, when the number of transmit antennas (M) is equal to the number of receive antennas (N), treating interference as noise performs the same as the HK scheme and is GDOF optimal. For K >; N/M+1, a combination of the HK and IA schemes performs the best among the schemes considered. However, for N/M <; K ≤ N/M+1, the HK scheme is found to be GDOF optimal.

On the capacity of quantized gaussian MAC channels with finite input alphabet

In this paper, we investigate the achievable rate region of Gaussian multiple access channels (MAC) with finite input alphabet and quantized output. With finite input alphabet and an unquantized receiver, the two-user Gaussian MAC rate region was studied. In most high throughput communication systems based on digital signal processing, the analog received signal is quantized using a low precision quantizer. In this paper, we first derive the expressions for the achievable rate region of a two-user Gaussian MAC with finite input alphabet and quantized output. We show that, with finite input alphabet, the achievable rate region with the commonly used uniform receiver quantizer has a significant loss in the rate region compared. It is observed that this degradation is due to the fact that the received analog signal is densely distributed around the origin, and is therefore not efficiently quantized with a uniform quantizer which has equally spaced quantization intervals. It is also observed that the density of the received analog signal around the origin increases with increasing number of users. Hence, the loss in the achievable rate region due to uniform receiver quantization is expected to increase with increasing number of users. We, therefore, propose a novel non-uniform quantizer with finely spaced quantization intervals near the origin. For a two-user Gaussian MAC with a given finite input alphabet and low precision receiver quantization, we show that the proposed non-uniform quantizer has a significantly larger rate region compared to what is achieved with a uniform quantizer.

A Huber-loss-driven clustering technique and its application to robust cell detection in confocal microscopy images

We address the problem of detecting cells in biological images. The problem is important in many automated image analysis applications. We identify the problem as one of clustering and formulate it within the framework of robust estimation using loss functions. We show how suitable loss functions may be chosen based on a priori knowledge of the noise distribution. Specifically, in the context of biological images, since the measurement noise is not Gaussian, quadratic loss functions yield suboptimal results. We show that by incorporating the Huber loss function, cells can be detected robustly and accurately. To initialize the algorithm, we also propose a seed selection approach. Simulation results show that Huber loss exhibits better performance compared with some standard loss functions. We also provide experimental results on confocal images of yeast cells. The proposed technique exhibits good detection performance even when the signal-to-noise ratio is low.

Mapping joint ROM on a cube using electromagnetic trackers

This paper presents a unified framework using the unit cube for measurement, representation and usage of the range of motion (ROM) of body joints with multiple degrees of freedom (d.o.f) to be used for digital human models (DHM). Traditional goniometry needs skill and kn owledge; it is intrusive and has limited applicability for multi-d.o.f. joints. Measurements using motion capture systems often involve complicated mathematics which itself need validation. In this paper we use change of orientation as the measure of rotation; this definition does not require the identification of any fixed axis of rotation. A two-d.o.f. joint ROM can be represented as a Gaussian map. Spherical polygon representation of ROM, though popular, remains inaccurate, vulnerable due to singularities on parametric sphere and difficult to use for point classification. The unit cube representation overcomes these difficulties. In the work presented here, electromagnetic trackers have been effectively used for measuring the relative orientation of a body segment of interest with respect to another body segment. The orientation is then mapped on a surface gridded cube. As the body segment is moved, the grid cells visited are identified and visualized. Using the visual display as a feedback, the subject is instructed to cover as many grid cells as he can. In this way we get a connected patch of contiguous grid cells. The boundary of this patch represents the active ROM of the concerned joint. The tracker data is converted into the motion of a direction aligned with the axis of the segment and a rotation about this axis later on. The direction identifies the grid cells on the cube and rotation about the axis is represented as a range and visualized using color codes. Thus the present methodology provides a simple, intuitive and accura te determination and representation of up to 3 d.o.f. joints. Basic results are presented for the shoulder. The measurement scheme to be used for wrist and neck, and approach for estimation of the statistical distribution of ROM for a given population are also discussed.

Narrowing of resonances in electromagnetically induced transparency and absorption using a Laguerre-Gaussian control beam

We study the phenomenon of electromagnetically induced transparency and absorption (EITA) using a control laser with a Laguerre-Gaussian (LG) profile instead of the usual Gaussian profile, and observe significant narrowing of the resonance widths. Aligning the probe beam to the central hole in the doughnut-shaped LG control beam allows simultaneously a strong control intensity required for high signal-to-noise ratio and a low intensity in the probe region required to get narrow resonances. Experiments with an expanded Gaussian control and a second-order LG control show that transit time and orbital angular momentum do not play a significant role. This explanation is borne out by a density-matrix analysis with a radially varying control Rabi frequency. We observe these resonances using degenerate two-level transitions in the D-2 line of Rb-87 in a room temperature vapor cell, and an EIA resonance with width up to 20 times below the natural linewidth for the F = 2 -> F' = 3 transition. Thus the use of LG beams should prove advantageous in all applications of EITA and other kinds of pump-probe spectroscopy as well.

Feasibility of laser-induced breakdown spectroscopy for the study of the temporal distribution of trace elements trapped in snow collected from greater himalayan range

The potential merit of laser-induced breakdown spectroscopy (LIBS) has been demonstrated for detection and quantification of trace pollutants trapped in snow/ice samples. In this technique, a high-power pulsed laser beam from Nd:YAG Laser (Model no. Surelite III-10, Continuum, Santa Clara, CA, USA) is focused on the surface of the target to generate plasma. The characteristic emissions from laser-generated plasma are collected and recorded by a fiber-coupled LIBS 2000+ (Ocean Optics, Santa Clara, CA, USA) spectrometer. The fingerprint of the constituents present in the sample is obtained by analyzing the spectral lines by using OOI LIBS software. Reliable detection of several elements like Zn, Al, Mg, Fe, Ca, C, N, H, and O in snow/ice samples collected from different locations (elevation) of Manali and several snow samples collected from the Greater Himalayan region (from a cold lab in Manali, India) in different months has been demonstrated. The calibration curve approach has been adopted for the quantitative analysis of these elements like Zn, Al, Fe, and Mg. Our results clearly demonstrate that the level of contamination is higher in those samples that were collected in the month of January in comparison to those collected in February and March.

A long-lost relic from the eastern ghats: morphology, distribution and habitat of sepsophis punctatus Beddome, 1870 (Squamata: Scincidae)

Sepsophis punctatus Beddome 1870, the only species of a monotypic genus, was described based on a single specimen from the Eastern Ghats of India. We rediscovered the species based on specimens from Odisha and Andhra Pradesh state, India, after a gap of 137 years, including four specimens from close to the type locality. The holotype was studied in detail, and we present additional morphological characters of the species with details on natural history, habitat and diet. The morphological characters of the holotype along with two additional specimens collected by Beddome are compared with the specimens collected by us. We also briefly discuss the distribution of other members of the subfamily Scincinae and their evolutionary affinities.

Power allocation for interference channels

We propose power allocation algorithms for increasing the sum rate of two and three user interference channels. The channels experience fast fading and there is an average power constraint on each transmitter. Our achievable strategies for two and three user interference channels are based on the classification of the interference into very strong, strong and weak interferences. We present numerical results of the power allocation algorithm for two user Gaussian interference channel with Rician fading with mean total power gain of the fade Omega = 3 and Rician factor kappa = 0.5 and compare the sum rate with that obtained from ergodic interference alignment with water-filling. We show that our power allocation algorithm increases the sum rate with a gain of 1.66dB at average transmit SNR of 5dB. For the three user Gaussian interference channel with Rayleigh fading with distribution CN(0, 0.5), we show that our power allocation algorithm improves the sum rate with a gain of 1.5dB at average transmit SNR of 5dB.

Capacity bounds for the Gaussian X channel

We consider bounds for the capacity region of the Gaussian X channel (XC), a system consisting of two transmit-receive pairs, where each transmitter communicates with both the receivers. We first classify the XC into two classes, the strong XC and the mixed XC. In the strong XC, either the direct channels are stronger than the cross channels or vice-versa, whereas in the mixed XC, one of the direct channels is stronger than the corresponding cross channel and vice-versa. After this classification, we give outer bounds on the capacity region for each of the two classes. This is based on the idea that when one of the messages is eliminated from the XC, the rate region of the remaining three messages are enlarged. We make use of the Z channel, a system obtained by eliminating one message and its corresponding channel from the X channel, to bound the rate region of the remaining messages. The outer bound to the rate region of the remaining messages defines a subspace in R-+(4) and forms an outer bound to the capacity region of the XC. Thus, the outer bound to the capacity region of the XC is obtained as the intersection of the outer bounds to the four combinations of the rate triplets of the XC. Using these outer bounds on the capacity region of the XC, we derive new sum-rate outer bounds for both strong and mixed Gaussian XCs and compare them with those existing in literature. We show that the sum-rate outer bound for strong XC gives the sum-rate capacity in three out of the four sub-regions of the strong Gaussian XC capacity region. In case of mixed Gaussian XC, we recover the recent results in 11] which showed that the sum-rate capacity is achieved in two out of the three sub-regions of the mixed XC capacity region and give a simple alternate proof of the same.

Bounds on the sum-rate capacity of the Gaussian MIMO X channel

We consider the MIMO X channel (XC), a system consisting of two transmit-receive pairs, where each transmitter communicates with both the receivers. Both the transmitters and receivers are equipped with multiple antennas. First, we derive an upper bound on the sum-rate capacity of the MIMO XC under individual power constraint at each transmitter. The sum-rate capacity of the two-user multiple access channel (MAC) that results when receiver cooperation is assumed forms an upper bound on the sum-rate capacity of the MIMO XC. We tighten this bound by considering noise correlation between the receivers and deriving the worst noise covariance matrix. It is shown that the worst noise covariance matrix is a saddle-point of a zero-sum, two-player convex-concave game, which is solved through a primal-dual interior point method that solves the maximization and the minimization parts of the problem simultaneously. Next, we propose an achievable scheme which employs dirty paper coding at the transmitters and successive decoding at the receivers. We show that the derived upper bound is close to the achievable region of the proposed scheme at low to medium SNRs.

Elasticity of DNA and the effect of dendrimer binding

Negatively charged DNA can be compacted by positively charged dendrimers and the degree of compaction is a delicate balance between the strength of the electrostatic interaction and the elasticity of DNA. We report various elastic properties of short double-stranded DNA (dsDNA) and the effect of dendrimer binding using fully atomistic molecular dynamics and numerical simulations. In equilibrium at room temperature, the contour length distribution P(L) and the end-to-end distance distribution P(R) are nearly Gaussian, the former gives an estimate of the stretch modulus gamma(1) of dsDNA in quantitative agreement with the literature value. The bend angle distribution P(.) of the dsDNA also has a Gaussian form and allows to extract a persistence length, L-p of 43 nm. When the dsDNA is compacted by positively charged dendrimer, the stretch modulus stays invariant but the effective bending rigidity estimated from the end-to-end distance distribution decreases dramatically due to backbone charge neutralization of dsDNA by dendrimer. We support our observations with numerical solutions of the worm-like-chain (WLC) model as well as using non-equilibrium dsDNA stretching simulations. These results are helpful in understanding the dsDNA elasticity at short length scales as well as how the elasticity is modulated when dsDNA binds to a charged object such as a dendrimer or protein.