Detecting rare events using Kullback-Leibler divergence: A weakly supervised approach

Video surveillance infrastructure has been widely installed in public places for security purposes. However, live video feeds are typically monitored by human staff, making the detection of important events as they occur difficult. As such, an expert system that can automatically detect events of interest in surveillance footage is highly desirable. Although a number of approaches have been proposed, they have significant limitations: supervised approaches, which can detect a specific event, ideally require a large number of samples with the event spatially and temporally localised; while unsupervised approaches, which do not require this demanding annotation, can only detect whether an event is abnormal and not specific event types. To overcome these problems, we formulate a weakly-supervised approach using Kullback-Leibler (KL) divergence to detect rare events. The proposed approach leverages the sparse nature of the target events to its advantage, and we show that this data imbalance guarantees the existence of a decision boundary to separate samples that contain the target event from those that do not. This trait, combined with the coarse annotation used by weakly supervised learning (that only indicates approximately when an event occurs), greatly reduces the annotation burden while retaining the ability to detect specific events. Furthermore, the proposed classifier requires only a decision threshold, simplifying its use compared to other weakly supervised approaches. We show that the proposed approach outperforms state-of-the-art methods on a popular real-world traffic surveillance dataset, while preserving real time performance.

Intra-band electron pairing in the presence of a radiation field

A semiconductor with almost overlapping conduction bands b and c is considered. It is found that an attractive interaction leading to superconductivity can be induced between electrons in the conduction band b by a strong radiation field of monochromatic photons whose energy differs slightly from the band gap Ebc. The mechanism is the exchange of a photon and a phonon between the interacting electrons and the interaction is found to be proportional to the photon density.

Photon-Induced Electron Pairing

The earlier work on the possibility of interband electron pairing in the presence of a strong radiation field has been further extended. Some additional terms, neglected earlier, have been taken into account and generalized to a situation where the electron-phonon coupling coefficients for the two conduction bands (valleys) are different. It is found that the pairing interaction is attractive and the strength depends on the photon density.

Shape-Dependent Confinement in Ultrasmall Zero-, One-, and Two-Dimensional PbS Nanostructures

Spatial dimensionality affects the degree of confinement when an electron-hole pair is squeezed from one or more dimensions approaching the bulk exciton Bohr radius (alpha(B)) limit. The etectron-hole interaction in zero-dimensional (0D) dots, one-dimensional (1D) rods/wires, and two-dimensional (2D) wells/sheets should be enhanced by the increase in confinement dimensions in the order 0D > 1D > 2D. We report the controlled synthesis of PbS nanomateriats with 0D, 1D, and 2D forms retaining at least one dimension in the strongly confined regime far below alpha(B) (similar to 10 nm for PbS) and provide evidence through varying the exciton-phonon coupling strength that the degree of confinement is systematically weakened by the loss of confinement dimension. Geometry variations show distinguishable far-field optical polarizations, which could find useful applications in polarization-sensitive devices.

Spatial and temporal variability in midge (Nematocera) assemblages in shallow Finnish lakes (60−70 °N) : community-based modelling of past environmental change

Multi- and intralake datasets of fossil midge assemblages in surface sediments of small shallow lakes in Finland were studied to determine the most important environmental factors explaining trends in midge distribution and abundance. The aim was to develop palaeoenvironmental calibration models for the most important environmental variables for the purpose of reconstructing past environmental conditions. The developed models were applied to three high-resolution fossil midge stratigraphies from southern and eastern Finland to interpret environmental variability over the past 2000 years, with special focus on the Medieval Climate Anomaly (MCA), the Little Ice Age (LIA) and recent anthropogenic changes. The midge-based results were compared with physical properties of the sediment, historical evidence and environmental reconstructions based on diatoms (Bacillariophyta), cladocerans (Crustacea: Cladocera) and tree rings. The results showed that the most important environmental factor controlling midge distribution and abundance along a latitudinal gradient in Finland was the mean July air temperature (TJul). However, when the dataset was environmentally screened to include only pristine lakes, water depth at the sampling site became more important. Furthermore, when the dataset was geographically scaled to southern Finland, hypolimnetic oxygen conditions became the dominant environmental factor. The results from an intralake dataset from eastern Finland showed that the most important environmental factors controlling midge distribution within a lake basin were river contribution, water depth and submerged vegetation patterns. In addition, the results of the intralake dataset showed that the fossil midge assemblages represent fauna that lived in close proximity to the sampling sites, thus enabling the exploration of within-lake gradients in midge assemblages. Importantly, this within-lake heterogeneity in midge assemblages may have effects on midge-based temperature estimations, because samples taken from the deepest point of a lake basin may infer considerably colder temperatures than expected, as shown by the present test results. Therefore, it is suggested here that the samples in fossil midge studies involving shallow boreal lakes should be taken from the sublittoral, where the assemblages are most representative of the whole lake fauna. Transfer functions between midge assemblages and the environmental forcing factors that were significantly related with the assemblages, including mean air TJul, water depth, hypolimnetic oxygen, stream flow and distance to littoral vegetation, were developed using weighted averaging (WA) and weighted averaging-partial least squares (WA-PLS) techniques, which outperformed all the other tested numerical approaches. Application of the models in downcore studies showed mostly consistent trends. Based on the present results, which agreed with previous studies and historical evidence, the Medieval Climate Anomaly between ca. 800 and 1300 AD in eastern Finland was characterized by warm temperature conditions and dry summers, but probably humid winters. The Little Ice Age (LIA) prevailed in southern Finland from ca. 1550 to 1850 AD, with the coldest conditions occurring at ca. 1700 AD, whereas in eastern Finland the cold conditions prevailed over a longer time period, from ca. 1300 until 1900 AD. The recent climatic warming was clearly represented in all of the temperature reconstructions. In the terms of long-term climatology, the present results provide support for the concept that the North Atlantic Oscillation (NAO) index has a positive correlation with winter precipitation and annual temperature and a negative correlation with summer precipitation in eastern Finland. In general, the results indicate a relatively warm climate with dry summers but snowy winters during the MCA and a cool climate with rainy summers and dry winters during the LIA. The results of the present reconstructions and the forthcoming applications of the models can be used in assessments of long-term environmental dynamics to refine the understanding of past environmental reference conditions and natural variability required by environmental scientists, ecologists and policy makers to make decisions concerning the presently occurring global, regional and local changes. The developed midge-based models for temperature, hypolimnetic oxygen, water depth, littoral vegetation shift and stream flow, presented in this thesis, are open for scientific use on request.

On the Geometry of Infinite-Dimensional Grassmannian Manifolds and Gauge Theory

This PhD Thesis is about certain infinite-dimensional Grassmannian manifolds that arise naturally in geometry, representation theory and mathematical physics. From the physics point of view one encounters these infinite-dimensional manifolds when trying to understand the second quantization of fermions. The many particle Hilbert space of the second quantized fermions is called the fermionic Fock space. A typical element of the fermionic Fock space can be thought to be a linear combination of the configurations m particles and n anti-particles . Geometrically the fermionic Fock space can be constructed as holomorphic sections of a certain (dual)determinant line bundle lying over the so called restricted Grassmannian manifold, which is a typical example of an infinite-dimensional Grassmannian manifold one encounters in QFT. The construction should be compared with its well-known finite-dimensional analogue, where one realizes an exterior power of a finite-dimensional vector space as the space of holomorphic sections of a determinant line bundle lying over a finite-dimensional Grassmannian manifold. The connection with infinite-dimensional representation theory stems from the fact that the restricted Grassmannian manifold is an infinite-dimensional homogeneous (Kähler) manifold, i.e. it is of the form G/H where G is a certain infinite-dimensional Lie group and H its subgroup. A central extension of G acts on the total space of the dual determinant line bundle and also on the space its holomorphic sections; thus G admits a (projective) representation on the fermionic Fock space. This construction also induces the so called basic representation for loop groups (of compact groups), which in turn are vitally important in string theory / conformal field theory. The Thesis consists of three chapters: the first chapter is an introduction to the backround material and the other two chapters are individually written research articles. The first article deals in a new way with the well-known question in Yang-Mills theory, when can one lift the action of the gauge transformation group on the space of connection one forms to the total space of the Fock bundle in a compatible way with the second quantized Dirac operator. In general there is an obstruction to this (called the Mickelsson-Faddeev anomaly) and various geometric interpretations for this anomaly, using such things as group extensions and bundle gerbes, have been given earlier. In this work we give a new geometric interpretation for the Faddeev-Mickelsson anomaly in terms of differentiable gerbes (certain sheaves of categories) and central extensions of Lie groupoids. The second research article deals with the question how to define a Dirac-like operator on the restricted Grassmannian manifold, which is an infinite-dimensional space and hence not in the landscape of standard Dirac operator theory. The construction relies heavily on infinite-dimensional representation theory and one of the most technically demanding challenges is to be able to introduce proper normal orderings for certain infinite sums of operators in such a way that all divergences will disappear and the infinite sum will make sense as a well-defined operator acting on a suitable Hilbert space of spinors. This research article was motivated by a more extensive ongoing project to construct twisted K-theory classes in Yang-Mills theory via a Dirac-like operator on the restricted Grassmannian manifold.

Infrared absorption spectra of As-Se glasses

IR absorption spectra of As-Se glasses have been studied over a wide range of compositions. Various two-phonon, multiphonon (combination tones) and impurity absorptions have been identified. Compositional variation of relative band intensities has been explained in terms of the chemically ordered network model.

Energy losses at pipe trifurcations

In the case of pipe trifurcation, previous observations report negative energy losses in the centre branch. This causes an anomaly, because there should not be any negative energy loss due to conservation of energy principle. Earlier investigators have suggested that this may be due to the non-inclusion of kinetic energy coefficient (a) in the computations of energy losses without any experimental evidence. In the present work, through experimentally determined velocity profiles, energy loss coefficients have been evaluated. It has been found that with the inclusion of a in the computations of energy loss, there is no negative energy loss in the centre branch.

Brillouin Scattering by Phonons Renormalized by Two-Magnon Excitations in Ferromagnetic Dielectrics

Brillouin scattering by one-phonon-two-magnon interacting excitations in ferromagnetic dielectrics is discussed. The basic light scattering mechanism is taken to be the modulation of the density-dependent optical dielectric polarizability of the medium by the dynamic strain field generated by the longitudinal acoustic (LA) phonons. The renormalization effects arising from the scattering of phonons by the two-magnon creation-annihilation processes have, however, been taken into account. Via these interactions, the Brillouin components corresponding to the two-magnon excitations are reflected indirectly in the spectrum of the phonon scattered light as line-broadening of the otherwise relatively sharp Brillouin doublet. The present mechanism is shown to be dominant in a clean saturated ferromagnetic dielectric with large magneto-strictive coupling constant, and with the magnetic ions in the orbitally quenched states. Following the linear response theory, an expression has been derived for the spectral density of the scattered light as a function of temperature, scattering angle, and the strength of the externally applied magnetic field. Some estimates are given for the line-width and line-shift of the Brillouin components for certain typical choice of parameters involved. The results are discussed in relation to some available calculations on the ultrasonic attenuation in ferromagnetic insulators at low temperatures.

The heat capacity of liquid carbon disulfide + acetonitrile, especially near the upper critical solution temperature

The heat capacity Cp of the binary liquid system CS2 + CH3CN has been studied. This system has an upper critical solution temperature To ≈ 323.4 K and a critical mole fraction of CS2xo ≈ 0.5920. Measurements were made both for mixtures close to and far away from the critical region. The heat capacity of the mixture with x = xo exhibits a symmetric logarithmic anomaly around Tc, which is apparently preserved even for compositions in the immediate vicinity of xc. For compositions far away from xc, only a normal rise in Cp over the covered temperature range is observed.

A Raman study of multiferroic LuMnO3

Magnetic and dielectric measurements confirm the multiferroic nature of LuMnO3. Raman spectra of LuMnO3 have been recorded in the 77-800 K range covering both the antiferromagnetic transition at 90 K and the ferroelectric-paraelectric transition at 750 K. The changes in the phonon modes frequencies and band-widths indicate the presence of phonon-spin coupling in the antiferromagnetically ordered phase. The ferroelectric-paraelectric transition is accompanied by the broadening and disappearance of many of the phonon modes. Some of the phonon modes also show anomalies at the ferroelectric transition.

Radiation induced enhancement of superconductivity

The effect of microwave radiation on the electron-phonon vertex in superconductors is taken into account. This leads to an enhancement of effective pairing interaction and hence to the transition temperature (Tc) which depends on the photon density and the frequency. This prediction is in agreement with recent experimental results.

Liquid-liquid critical phenomena. The coexistence curve of n-heptane-acetic anhydride

The coexistence curve of the binary liquid mixture n-heptane-acetic anhydride has been determined by the observation of the transition temperatures of 76 samples over the range of compositions. The functional form of the difference in order parameter, in terms of either the mole fraction or the volume fraction, is consistent with theoretical predictions invoking the concept of universality at critical points. The average value of the order parameter, the diameter of the coexistence curve, shows an anomaly which can be described by either an exponent 1 - a, as predicted by various theories (where a is the critical exponent of the specific heat), or by an exponent 20 (where P is the coexistence curve exponent), as expected when the order parameter used is not the one the diameter of which diverges asymptotically as 1 - a.

R1STM: One-class support tensor machine with randomised kernel

Identifying unusual or anomalous patterns in an underlying dataset is an important but challenging task in many applications. The focus of the unsupervised anomaly detection literature has mostly been on vectorised data. However, many applications are more naturally described using higher-order tensor representations. Approaches that vectorise tensorial data can destroy the structural information encoded in the high-dimensional space, and lead to the problem of the curse of dimensionality. In this paper we present the first unsupervised tensorial anomaly detection method, along with a randomised version of our method. Our anomaly detection method, the One-class Support Tensor Machine (1STM), is a generalisation of conventional one-class Support Vector Machines to higher-order spaces. 1STM preserves the multiway structure of tensor data, while achieving significant improvement in accuracy and efficiency over conventional vectorised methods. We then leverage the theory of nonlinear random projections to propose the Randomised 1STM (R1STM). Our empirical analysis on several real and synthetic datasets shows that our R1STM algorithm delivers comparable or better accuracy to a state-of-the-art deep learning method and traditional kernelised approaches for anomaly detection, while being approximately 100 times faster in training and testing.

Femtosecond carrier dynamics and saturable absorption in graphene suspensions

Nonlinear optical properties and carrier relaxation dynamics in graphene, suspended in three different solvents, are investigated using femtosecond (80 fs pulses) Z-scan and degenerate pump-probe spectroscopy at 790 nm. The results demonstrate saturable absorption property of graphene with a nonlinear absorption coefficient, beta of (similar to 2-9) x 10(-8) cm/W. Two distinct time scales associated with the relaxation of photoexcited carriers, a fast one in the range of 130-330 fs (related to carrier-carrier scattering) followed by it slower one in 3.5-4.9 ps range (associated with carrier-phonon scattering) are observed. (C) 2009 American Institute of Physics.