We present topic-regression multi-modal Latent Dirich-let Allocation (tr-mmLDA), a novel statistical topic model for the task of image and video annotation. At the heart of our new annotation model lies a novel latent variable regression approach to capture correlations between image or video features and annotation texts. Instead of sharing a set of latent topics between the 2 data modalities as in the formulation of correspondence LDA in [2], our approach introduces a regression module to correlate the 2 sets of topics, which captures more general forms of association and allows the number of topics in the 2 data modalities to be different. We demonstrate the power of tr-mmLDA on 2 standard annotation datasets: a 5000-image subset of COREL and a 2687-image LabelMe dataset. The proposed association model shows improved performance over correspondence LDA as measured by caption perplexity.

This paper presents a new probabilistic model for the task of image annotation. Our model, which we call sLDA-bin, extends supervised Latent Dirichlet Allocation (sLDA) model to handle a multi-variate binary response variable of the annotation data. Unlike correspondence LDA (cLDA), the association model in sLDA allows each caption word to be associated with more than 1 image region and is thus more appropriate for annotation words that globally describe the scene. By modeling the response variable as a multi-variate Bernoulli, we introduce a tight convex variational bound for the logistic function and derive an efficient variational inference algorithm based on mean-field approximation. Our model compares favorably with cLDA on an image annotation task, as demonstrated by a superior caption prediction probability.

Topic models such as Latent Dirichlet Allocation (LDA) and Correlated Topic Model (CTM) have recently emerged as powerful statistical tools for text document modeling. In this paper, we improve upon CTM and propose Independent Factor Topic Models (IFTM) which use linear latent variable models to uncover the hidden sources of correlation between topics. There are 2 main contributions of this work. First, by using a sparse source prior model, we can directly visualize sparse patterns of topic correlations. Secondly, the conditional independence assumption implied in the use of latent source variables allows the objective function to factorize, leading to a fast Newton-Raphson based variational inference algorithm. Experimental results on synthetic and real data show that IFTM runs on average 3--5 times faster than CTM, while giving competitive performance as measured by perplexity and loglikelihood of held-out data.

This paper presents a new approximate Bayesian estimator for enhancing a noisy speech signal. The speech model is assumed to be a Gaussian mixture model (GMM) in the log-spectral domain. This is in contrast to most current models in frequency domain. Exact signal estimation is a computationally intractable problem. We derive three approximations to enhance the efficiency of signal estimation. The Gaussian approximation transforms the log-spectral domain GMM into the frequency domain using minimal Kullback-Leiber (KL)-divergency criterion. The frequency domain Laplace method computes the maximum a posteriori (MAP) estimator for the spectral amplitude. Correspondingly, the log-spectral domain Laplace method computes the MAP estimator for the log-spectral amplitude. Further, the gain and noise spectrum adaptation are implemented using the expectation-maximization (EM) algorithm within the GMM under Gaussian approximation. The proposed algorithms are evaluated by applying them to enhance the speeches corrupted by the speech-shaped noise (SSN). The experimental results demonstrate that the proposed algorithms offer improved signal-to-noise ratio, lower word recognition error rate, and less spectral distortion.

The synchronous brain activity measured via MEG (or EEG) can be interpreted as arising from a collection (possibly large) of current dipoles or sources located throughout the cortex. Estimating the number, location, and orientation of these sources remains a challenging task, one that is significantly compounded by the effects of source correlations and the presence of interference from spontaneous brain activity, sensor noise, and other artifacts. This paper derives an empirical Bayesian method for addressing each of these issues in a principled fashion. The resulting algorithm guarantees descent of a cost function uniquely designed to handle unknown orientations and arbitrary correlations. Robust interference suppression is also easily incorporated. In a restricted setting, the proposed method is shown to have theoretically zero bias estimating both the location and orientation of multi-component dipoles even in the presence of correlations, unlike a variety of existing Bayesian localization methods or common signal processing techniques such as beamforming and sLORETA. Empirical results on both simulated and real data sets verify the efficacy of this approach.

In this paper, we develop a maximum-likelihood (ML) spatio-temporal blind source separation (BSS) algorithm, where the temporal dependencies are explained by assuming that each source is an autoregressive (AR) process and the distribution of the associated independent identically distributed (i.i.d.) innovations process is described using a mixture of Gaussians. Unlike most ML methods, the proposed algorithm takes into account both spatial and temporal information, optimization is performed using the expectation-maximization (EM) method, the source model is adapted to maximize the likelihood, and the update equations have a simple, analytical form. The proposed method, which we refer to as autoregressive mixture of Gaussians (AR-MOG), outperforms nine other methods for artificial mixtures of real audio. We also show results for using AR-MOG to extract the fetal cardiac signal from real magnetocardiographic (MCG) data.

We present a probabilistic graphical model that learns the joint statistical structures of text, audio, and video for the purpose of classification and retrieval of multimedia documents. The proposed model, which we call multi-modal LDA (MM-LDA), builds on the basic latent Dirichlet allocation (LDA) model by postulating common hidden factors, termed topics, that are shared among the 3 data modalities. These hidden topics correspond to patterns of word co-occurrences in multimedia documents and describe how text words co-occur with certain visual and acoustic features. We demonstrate the power of MM-LDA in representing TV clips containing closed-captions, video, and audio, and show promising results in 3 challenging applications: TV clip classification, retrieval, and auto-annotation.

A novel Kalman filtering/smoothing algorithm is presented for efficient and accurate estimation of vocal tract resonances or formants, which are natural frequencies and bandwidths of the resonator from larynx to lips, in fluent speech. The algorithm uses a hidden dynamic model, with a state-space formulation, where the resonance frequency and bandwidth values are treated as continuous-valued hidden state variables. The observation equation of the model is constructed by an analytical predictive function from the resonance frequencies and bandwidths to LPC cepstra as the observation vectors. This nonlinear function is adaptively linearized, and a residual or bias term, which is adaptively trained, is added to the nonlinear function to represent the iteratively reduced piecewise linear approximation error. Details of the piecewise linearization design process are described. An iterative tracking algorithm is presented, which embeds both the adaptive residual training and piecewise linearization design in the Kalman filtering/smoothing framework. Experiments on estimating resonances in Switchboard speech data show accurate estimation results. In particular, the effectiveness of the adaptive residual training is demonstrated. Our approach provides a solution to the traditional "hidden formant problem," and produces meaningful results even during consonantal closures when the supra-laryngeal source may cause no spectral prominences in speech acoustics

Blind source separation (BSS) is a challenging problem in real-world environments where sources are time delayed and convolved. The problem becomes more difficult in very reverberant conditions, with an increasing number of sources, and geometric configurations of the sources such that finding directionality is not sufficient for source separation. In this paper, we propose a new algorithm that exploits higher order frequency dependencies of source signals in order to separate them when they are mixed. In the frequency domain, this formulation assumes that dependencies exist between frequency bins instead of defining independence for each frequency bin. In this manner, we can avoid the well-known frequency permutation problem. To derive the learning algorithm, we define a cost function, which is an extension of mutual information between multivariate random variables. By introducing a source prior that models the inherent frequency dependencies, we obtain a simple form of a multivariate score function. In experiments, we generate simulated data with various kinds of sources in various environments. We evaluate the performances and compare it with other well-known algorithms. The results show the proposed algorithm outperforms the others in most cases. The algorithm is also able to accurately recover six sources with six microphones. In this case, we can obtain about 16-dB signal-to-interference ratio (SIR) improvement. Similar performance is observed in real conference room recordings with three human speakers reading sentences and one loudspeaker playing music

We propose a novel approach to the blind source separation (BSS) that exploits frequency dependencies within a source. In contrast to conventional algorithms that separate the sources independently in each frequency bin, we assume that dependencies exist between frequency bins in a source signal. In this manner, we can reduce or eliminate the well-known frequency permutation problem. We derive the learning algorithm by defining a cost function as an extension of mutual information between multivariate random variables and by introducing a source prior that models the inherent frequency dependencies. This results in a simple form of a multivariate score function. In simulations and real recording experiments, we evaluate the performance of the proposed method and compare it against other well-known algorithms under various conditions. Our results indicate that modeling dependencies yields improved performance and robust scaling to higher number of sources and mixtures.

Voice based computing applications, such as phone communication and speech recognition, use microphone arrays to capture voice from a human speaker. In many environments of interest, however, sounds from other sources interfere with the speaker's voice, posing severe problems for subsequent processing. This paper describes a new framework for treating this problem, and presents and demonstrates a new algorithm for the cancellation of interfering sounds. Our framework combines techniques from statistical machine learning with ideas from speech and audio processing. An important feature involves training rich probabilistic models on data from different types of relevant sound sources. Those source models are then incorporated into a larger probabilistic model of the observed microphone data. Using that model we derive our algorithm, which is of the expectation-maximization type and infers from data the clean sound of separate individual sources. We report very good results on data recorded in different environments

We have developed a novel algorithm for integrating source localization and noise suppression based on a probabilistic graphical model of stimulus-evoked MEG/EEG data. Our algorithm localizes multiple dipoles while suppressing noise sources with the computational complexity equivalent to a single dipole scan, and is therefore more ef(cid:2)cient than traditional multidipole (cid:2)tting procedures. In simulation, the algorithm can accurately localize and estimate the time course of several simultaneously-active dipoles, with rotating or (cid:2)xed orientation, at noise levels typical for averaged MEG data. Furthermore, the algorithm is superior to beamforming techniques, which we show to be an approximation to our graphical model, in estimation of temporally correlated sources. Success of this algorithm for localizing auditory cortex in a tumor patient and for localizing an epileptic spike source are also demonstrated.

The paper presents a fast new algorithm for estimating the relative time delay between a microphone pair in noisy reverberant conditions. The algorithm is derived from a novel approach to this problem, which we formulate in the framework of probabilistic graphical models. We construct a probabilistic model of the microphone signals, using models of speech and noise as building blocks. The reverberation filter coefficients and the relative time delay appear as model parameters, and are estimated by our algorithm from data. The resulting delay estimate is Bayes optimal and takes into account noise and reverberation in a principled manner. We demonstrate very good performance on data from real and simulated room environments.

This paper presents a novel technique for analyzing electromagnetic imaging data obtained using the stimulus evoked experimental paradigm. The technique is based on a probabilistic graphical model, which describes the data in terms of underlying evoked and interference sources, and explicitly models the stimulus evoked paradigm. A variational Bayesian EM algorithm infers the model from data, suppresses interference sources, and reconstructs the activity of separated individual brain sources. The new algorithm outperforms existing techniques on two real datasets, as well as on simulated data.

A novel approach is developed for efficient and accurate tracking of vocal tract resonances, which are natural frequencies of the resonator from larynx to lips, in fluent speech. The tracking algorithm is based on a version of the structured speech model consisting of continuous-valued hidden dynamics and a piecewise-linearized prediction function from resonance frequencies and bandwidths to LPC cepstra. We present details of the piecewise linearization design process and an adaptive training technique for the parameters that characterize the prediction residuals. An iterative tracking algorithm is described and evaluated that embeds both the prediction-residual training and the piecewise linearization design in an adaptive Kalman filtering framework. Experiments on tracking vocal tract resonances in Switchboard speech data demonstrate high accuracy in the results, as well as the effectiveness of residual training embedded in the algorithm. Our approach differs from traditional formant trackers in that it provides meaningful results even during consonantal closures when the supra-laryngeal source may cause no spectral prominences in speech acoustics.

Perceiving sounds in a noisy environment is a challenging problem. Visual lip-reading can provide relevant information but is also challenging because lips are moving and a tracker must deal with a variety of conditions. Typically audio-visual systems have been assembled from individually engineered modules. We propose to fuse audio and video in a probabilistic generative model that implements cross-model self-supervised learning, enabling adaptation to audio-visual data. The video model features a Gaussian mixture model embedded in a linear subspace of a sprite which translates in the video. The system can learn to detect and enhance speech in noise given only a short (30 second) sequence of audio-visual data. We show some results for speech detection and enhancement, and discuss extensions to the model that are under investigation.

We present a method for separating two speakers from a single microphone channel. The method exploits the fine structure of male and female speech and relies on a strong high frequency resolution model for the source signals. The algorithm is able to identify the correct combination of male and female speech that best explains an observation and is able to reconstruct the component signals, relying on prior knowledge to 'fill in' regions that are masked by the other speaker. The two speaker single microphone source separation problem is one of the most challenging source separation scenarios and few quantitative results have been reported in the literature. We provide a test set based on the Aurora 2 data set and report performance numbers on a portion of this set. We achieve results of 6.59 dB average increase in SNR for female speakers and 5.51 dB for male speakers.

An important general model for discrete-time signal processing is the switching state space (SSS) model, which generalizes the hidden Markov model and the Gaussian state space model. Inference and parameter estimation in this model are known to be computationally intractable. This paper presents a powerful new approximation to the SSS model. The approximation is based on a variational technique that preserves the multimodal nature of the continuous state posterior distribution. Furthermore, by incorporating a windowing technique, the resulting EM algorithm has complexity that is just linear in the length of the time series. An alternative Viterbi decoding with frame-based likelihood is also presented which is crucial for the speech application that originally motivates this work. Our experiments focus on demonstrating the effectiveness of the algorithm by extensive simulations. A typical example in speech processing is also included to show the potential of this approach for practical applications.

This paper presents new algorithms for source separation with a microphone array. Key to our algorithms are exploiting detailed source models, using subband filtering ideas to model the reverberant environment, and employing explicit models for background and sensor noise. We demonstrate strong performance which is robust to noise and reverberations. Good scaling properties are obtained using machine learning techniques.

This paper describes novel and powerful variational EM algorithms for the segmental switching state space models used in speech applications, which are capable of capturing key internal (or hidden) dynamics of natural speech production. Hidden dynamic models (HDMs) have recently become a class of promising acoustic models to incorporate crucial speech-specific knowledge and overcome many inherent weaknesses of traditional HMMs. However, the lack of powerful and efficient statistical learning algorithms is one of the main obstacles preventing them from being well studied and widely used. Since exact inference and learning are intractable, a variational approach is taken to develop effective approximate algorithms. We have implemented the segmental constraint crucial for modeling speech dynamics and present algorithms for recovering hidden speech dynamics and discrete speech units from acoustic data only. The effectiveness of the algorithms developed are verified by experiments on simulation and Switchboard speech data.

We present a new approach to modeling and processing multimedia data. This approach is based on graphical models that combine audio and video variables. We demonstrate it by developing a new algorithm for tracking a moving object in a cluttered, noisy scene using two microphones and a camera. Our model uses unobserved variables to describe the data in terms of the process that generates them. It is therefore able to capture and exploit the statistical structure of the audio and video data separately, as well as their mutual dependencies. Model parameters are learned from data via an EM algorithm, and automatic calibration is performed as part of this procedure. Tracking is done by Bayesian inference of the object location from data. We demonstrate successful performance on multimedia clips captured in real world scenarios using off-the-shelf equipment.

We present a self-calibrating algorithm for audio-visual tracking using two microphones and a camera. The algorithm uses a parametrized statistical model which combines simple models of video and audio. Using unobserved variables, the model describes the process that generates the observed data. Hence, it is able to capture and exploit the statistical structure of the audio and video data, as well as their mutual dependencies, The model parameters are estimated by the EM algorithm; object templates are learned and automatic calibration is performed as part of this procedure. Tracking is done by Bayesian inference of the object location using the model. Successful performance is demonstrated on real multimedia clips.

Speech enhancement and recognition in noisy, reverberant conditions is a challenging open problem. We present a new approach to this problem, which is developed in the framework of probabilistic modeling. Our approach incorporates information about the statistical structure of speech signals using a speech model, which is pre-trained on a large dataset of clean speech. The speech model is a component in a larger model describing the observed sensor signals. That model is parametrized by the coefficients of the reverberation filters and the spectra of the sensor noise. We develop an EM algorithm that estimates those parameters from data and constructs a Bayes optimal estimator of the original speech signal. 1.

Source separation is an important problem at the intersection of several fields, including machine learning, signal processing, and speech tech- nology. Here we describe new separation algorithms which are based on probabilistic graphical models with latent variables. In contrast with existing methods, these algorithms exploit detailed models to describe source properties. They also use subband filtering ideas to model the reverberant environment, and employ an explicit model for background and sensor noise. We leverage variational techniques to keep the compu- tational complexity per EM iteration linear in the number of frames. 1 The Source Separation Problem Fig. 1 illustrates the problem of source separation with a sensor array. In this problem, signals from K independent sources are received by each of L (cid:21) K sensors. The task is to extract the sources from the sensor signals. It is a difficult task, partly because the received signals are distorted versions of the originals. There are two types of distortions. The first type arises from propagation through a medium, and is approximately linear but also history dependent. This type is usually termed reverberations. The second type arises from background noise and sensor noise, which are assumed additive. Hence, the actual task is to obtain an optimal estimate of the sources from data. The task is difficult for another reason, which is lack of advance knowledge of the properties of the sources, the propagation medium, and the noises. This difficulty gave rise to adaptive source separation algorithms, where parameters that are related to those properties are adjusted to optimized a chosen cost function. Unfortunately, the intense activity this problem has attracted over the last several years [1–9] has not yet produced a satisfactory solution. In our opinion, the reason is that existing tech- niques fail to address three major factors. The first is noise robustness: algorithms typically ignore background and sensor noise, sometime assuming they may be treated as additional sources. It seems plausible that to produce a noise robust algorithm, noise signals and their properties must be modeled explicitly, and these models should be exploited to compute optimal source estimators. The second factor is mixing filters: algorithms typically seek, and directly optimize, a transformation that would unmix the sources. However, in many situations, the filters describing medium propagation are non-invertible, or have an unstable inverse, or have a stable inverse that is extremely long. It may hence be advantageous to Figure 1: The source separation problem. Signals from K = 2 speakers propagate toward L = 2 sensors. Each sensor receives a linear mixture of the speaker signals, distorted by multipath propagation, medium response, and background and sensor noise. The task is to infer the original signals from sensor data. estimate the mixing filters themselves, then use them to estimate the sources. The third factor is source properties: algorithms typically use a very simple source model (e.g., a one time point histogram). But in many cases one may easily obtain detailed models of the source signals. This is particularly true for speech sources, where large datasets exist and much modeling expertise has developed over decades of research. Separation of speakers is also one of the major potential commercial applications of source separation algorithms. It seems plausible that incorporating strong source models could improve performance. Such models may potentially have two more advantages: first, they could help limit the range of possible mixing filters by constraining the optimization problem. Second, they could help avoid whitening the extracted signals by effectively limiting their spectral range to the range characteristic of the source model. This paper makes several contributions to the problem of real world source separation. In the following, we present new separation algorithms that are the first to address all three factors. We work in the framework of probabilistic graphical models. This framework allows us to construct models for sources and for noise, combine them with the reverberant mixing transformation in a principled manner, and compute parameter and source estimates from data which are Bayes optimal. We identify three technical ideas that are key to our approach: (1) a strong speech model, (2) subband filtering, and (3) variational EM. 2 Frames, Subband Signals, and Subband Filtering We start with the concept of subband filtering. This is also a good point to define our notation. Let xm denote a time domain signal, e.g., the value of a sound pressure waveform at time point m = 0; 1; 2; :::. Let Xn[k] denote the corresponding subband signal at time frame n and subband frequency k. The subband signals are obtained from the time domain signal by imposing an N-point window wm, m = 0 : N (cid:0) 1 on that signal at equally spaced points nJ, n = 0; 1; 2; :::, and FFT-ing the windowed signal,

We present a new method for speech denoising and robust speech recognition. Using the framework of probabilistic models allows us to integrate detailed speech models and models of realistic non-stationary noise signals in a principled manner. The framework transforms the denoising problem into a problem of Bayes-optimal signal estimation, producing minimum mean square error estimators of desired features of clean speech from noisy data. We describe a fast and efficient implementation of an algorithm that computes these estimators. The effectiveness of this algorithm is demonstrated in robust speech recognition experiments, using the Wall Street Journal speech corpus and Microsoft Whisper large-vocabulary continuous speech recognizer. Results show significantly lower word error rates than those under noisy-matched condition. In particular, when the denoising algorithm is applied to the noisy training data and subsequently the recognizer is retrained, very low error rates are obtained.

This paper presents a unified probabilistic framework for denoising and dereverberation of speech signals. The framework transforms the denois(cid:173) ing and dereverberation problems into Bayes-optimal signal estimation. The key idea is to use a strong speech model that is pre-trained on a large data set of clean speech. Computational efficiency is achieved by using variational EM, working in the frequency domain, and employing conjugate priors. The framework covers both single and multiple micro(cid:173) phones. We apply this approach to noisy reverberant speech signals and get results substantially better than standard methods.

We introduce the independent factor analysis (IFA) method for recovering independent hidden sources from their observed mixtures. IFA generalizes and unifies ordinary factor analysis (FA), principal component analysis (PCA), and independent component analysis (ICA), and can handle not only square noiseless mixing but also the general case where the number of mixtures differs from the number of sources and the data are noisy. IFA is a two-step procedure. In the first step, the source densities, mixing matrix, and noise covariance are estimated from the observed data by maximum likelihood. For this purpose we present an expectation-maximization (EM) algorithm, which performs unsupervised learning of an associated probabilistic model of the mixing situation. Each source in our model is described by a mixture of gaussians; thus, all the probabilistic calculations can be performed analytically. In the second step, the sources are reconstructed from the observed data by an optimal nonlinear estimator. A variational approximation of this algorithm is derived for cases with a large number of sources, where the exact algorithm becomes intractable. Our IFA algorithm reduces to the one for ordinary FA when the sources become gaussian, and to an EM algorithm for PCA in the zero-noise limit. We derive an additional EM algorithm specifically for noiseless IFA. This algorithm is shown to be superior to ICA since it can learn arbitrary source densities from the data. Beyond blind separation, IFA can be used for modeling multidimensional data by a highly constrained mixture of gaussians and as a tool for nonlinear signal encoding.

We derive a novel family of unsupervised learning algorithms for blind separation of mixed and convolved sources. Our approach is based on formulating the separation problem as a learning task of a spatiotemporal generative model, whose parameters are adapted iteratively to minimize suitable error functions, thus ensuring stability of the algorithms. The resulting learning rules achieve separation by exploiting high-order spatiotemporal statistics of the mixture data. Different rules are obtained by learning generative models in the frequency and time domains, whereas a hybrid frequency-time model leads to the best performance. These algorithms generalize independent component analysis to the case of convolutive mixtures and exhibit superior performance on instantaneous mixtures. An extension of the relative-gradient concept to the spatiotemporal case leads to fast and efficient learning rules with equivariant properties. Our approach can incorporate information about the mixing situation when available, resulting in a “semiblind” separation method. The spatiotemporal redundancy reduction performed by our algorithms is shown to be equivalent to information-rate maximization through a simple network. We illustrate the performance of these algorithms by successfully separating instantaneous and convolutive mixtures of speech and noise signals.