"Neurophysiology is increasingly focused on identifying coincident activity among neurons. Strong inferences about neural computation are made from the results of such studies, so it is important that these results be accurate. However, the preliminary step in the analysis of such data, the assignment of spike waveforms to individual neurons (“spike-sorting”), makes a critical assumption which undermines the analysis: that spikes, and hence neurons, are independent. We show that this assumption guarantees that coincident spiking estimates such as correlation coefficients are biased. We also show how to eliminate this bias. Our solution involves sorting spikes jointly, which contrasts with the current practice of sorting spikes independently of other spikes. This new “ensemble sorting” yields unbiased estimates of coincident spiking, and permits more data to be analyzed with confidence, improving the quality and quantity of neurophysiological inferences. These results should be of interest outside the context of neuronal correlations studies. Indeed, simultaneous recording of many neurons has become the rule rather than the exception in experiments, so it is essential to spike sort correctly if we are to make valid inferences about any properties of, and relationships between, neurons."

"We consider the sparse inverse covariance regularization problem or graphical lasso with regularization parameter λ. Suppose the sample covariance graph formed by thresholding the entries of the sample covariance matrix at λ is decomposed into connected components. We show that the vertex-partition induced by the connected components of the thresholded sample covariance graph (at λ) is exactly equal to that induced by the connected components of the estimated concentration graph, obtained by solving the graphical lasso problem for the same λ. This characterizes a very interesting property of a path of graphical lasso solutions. Furthermore, this simple rule, when used as a wrapper around existing algorithms for the graphical lasso, leads to enormous performance gains. For a range of values of λ, our proposal splits a large graphical lasso problem into smaller tractable problems, making it possible to solve an otherwise infeasible large-scale problem. We illustrate the graceful scalability of our proposal via synthetic and real-life microarray examples."
--- I wonder whether this hasn't some application to the PC algorithm?

"The Lasso is a popular model selection and estimation procedure for linear models that enjoys nice theoretical properties. In this paper, we study the Lasso estimator for fitting autoregressive time series models. We adopt a double asymptotic framework where the maximal lag may increase with the sample size. We derive theoretical results establishing various types of consistency. In particular, we derive conditions under which the Lasso estimator for the autoregressive coefficients is model selection consistent, estimation consistent and prediction consistent. Simulation study results are reported."

"Information estimates such as the ``direct method'' of Strong et al. (1998) sidestep the difficult problem of estimating the joint distribution of response and stimulus by instead estimating the difference between the marginal and conditional entropies of the response. While this is an effective estimation strategy, it tempts the practitioner to ignore the role of the stimulus and the meaning of mutual information. We show here that, as the number of trials increases indefinitely, the direct (or ``plug-in'') estimate of marginal entropy converges (with probability 1) to the entropy of the time-averaged conditional distribution of the response, and the direct estimate of the conditional entropy converges to the time-averaged entropy of the conditional distribution of the response. Under joint stationarity and ergodicity of the response and stimulus, the difference of these quantities converges to the mutual information. When the stimulus is deterministic or non-stationary the direct estimate of information no longer estimates mutual information, which is no longer meaningful, but it remains a measure of variability of the response distribution across time."

"We study a mechanism for reliable switching in biomolecular signal-transduction cascades. Steady bistable states are created by system-size cooperative effects in populations of proteins, in spite of the fact that the phosphorylation-state transitions of any molecule, by means of which the switch is implemented, are highly stochastic. The emergence of switching is a nonequilibrium phase transition in an energetically driven, dissipative system described by a master equation. We use operator and functional integral methods from reaction-diffusion theory to solve for the phase structure, noise spectrum, and escape trajectories and first-passage times of a class of minimal models of switches, showing how all critical properties for switch behavior can be computed within a unified framework."

"We propose a Bayesian nonparametric approach to the problem of jointly modeling multiple related time series. Our approach is based on the discovery of a set of latent, shared dynamical behaviors. Using a beta process prior, the size of the set and the sharing pattern are both inferred from data. We develop efficient Markov chain Monte Carlo methods based on the Indian buffet process representation of the predictive distribution of the beta process, without relying on a truncated model. In particular, our approach uses the sum-product algorithm to efficiently compute Metropolis-Hastings acceptance probabilities, and explores new dynamical behaviors via birth and death proposals. We examine the benefits of our proposed feature-based model on several synthetic datasets, and also demonstrate promising results on unsupervised segmentation of visual motion capture data."

"We present a framework for performing efficient regression in general metric spaces. Roughly speaking, our regressor predicts the value at a new point by computing a Lipschitz extension -- the smoothest function consistent with the observed data -- while performing an optimized structural risk minimization to avoid overfitting. The offline (learning) and online (inference) stages can be solved by convex programming, but this naive approach has runtime complexity O(n^3), which is prohibitive for large datasets. We design instead an algorithm that is fast when the the doubling dimension, which measures the "intrinsic" dimensionality of the metric space, is low.
We use the doubling dimension multiple times; first, on the statistical front, to bound fat-shattering dimension of the class of Lipschitz functions (and obtain risk bounds); and second, on the computational front, to quickly compute a hypothesis function and a prediction based on Lipschitz extension. Our resulting regressor is both asymptotically strongly consistent and comes with finite-sample risk bounds, while making minimal structural and noise assumptions."

Look at this later and re-consider the to_teach tags.

Heard the talk at ISIT 2011; it was very good.

Actually, I heard the talk about this and some even more impressive follow-up work, which Jadbabaie said will be in working paper form very soon.

"We investigate approximating joint distributions of random processes with causal dependence tree distributions. Such distributions are particularly useful in providing parsimonious representation when there exists causal dynamics among processes. By extending the results by Chow and Liu on dependence tree approximations, we show that the best causal dependence tree approximation is the one which maximizes the sum of directed informations on its edges, where best is defined in terms of minimizing the KL-divergence between the original and the approximate distribution. Moreover, we describe a low-complexity algorithm to efficiently pick this approximate distribution."

"A novel paradigm was developed to study the behavior of groups of networked people searching a problem space. The authors examined how different network structures affect the propagation of information in laboratory-created groups. Participants made numerical guesses and received scores that were also made available to their neighbors in the network. The networks were compared on speed of discovery and convergence on the optimal solution. One experiment showed that individuals within a group tend to converge on similar solutions even when there is an equally valid alternative solution. Two additional studies demonstrated that the optimal network structure depends on the problem space being explored, with networks that incorporate spatially based cliques having an advantage for problems that benefit from broad exploration, and networks with greater long-range connectivity having an advantage for problems requiring less exploration."

Heard the talk in Bristol. Naturally the idea makes a lot more sense when laid out by a proponent than by a hostile critic; it's actually very similar to indirect inference (only with a prior to add bias).

"general problem of adaptively controlling and/or identifying a stochastic dynamical system, where our {\em a priori} knowledge allows us to place the system in a subset of a metric space (the uncertainty set). We present an information-theoretic meta-theorem that captures the trade-off between the metric complexity (or richness) of the uncertainty set, the amount of information acquired online in the process of controlling and observing the system, and the residual uncertainty remaining after the observations have been collected. Following the approach of Zames, we quantify {\em a priori} information by the Kolmogorov (metric) entropy of the uncertainty set, while the information acquired online is expressed as a sum of information divergences. The general theory is used to derive new minimax lower bounds on the metric identification error, as well as to give a simple derivation of the minimum time needed to stabilize an uncertain stochastic linear system."

And by "heard the talk" I mean "had it explained on napkin-back over beer".

"A model for network panel data is discussed, based on the assumption that the observed data are discrete observations of a continuous-time Markov process on the space of all directed graphs on a given node set, in which changes in tie variables are independent conditional on the current graph. The model for tie changes is parametric and designed for applications to social network analysis, where the network dynamics can be interpreted as being generated by choices made by the social actors represented by the nodes of the graph. An algorithm for calculating the Maximum Likelihood estimator is presented, based on data augmentation and stochastic approximation. An application to an evolving friendship network is given and a small simulation study is presented which suggests that for small data sets the Maximum Likelihood estimator is more efficient than the earlier proposed Method of Moments estimator."

Best Student Paper at UAI 2010. What would happen if you used this on sequences of values from the Arnold cat map? Could it learn the direction of time?

Social influence on diagnosis, not actually producing autism. Heard the talk at the MERSI conference in 2009; it sounded pretty convincing.

"In a variety of disciplines such as social sciences, psychology, medicine and economics, the recorded data are considered to be noisy measurements of latent variables connected by some causal structure. This corresponds to a family of graphical models known as the structural equation model with latent variables. While linear non-Gaussian variants have been well-studied, inference in nonparametric structural equation models is still underdeveloped. We introduce a sparse Gaussian process parameterization that defines a non-linear structure connecting latent variables, unlike common formulations of Gaussian process latent variable models. An efficient Markov chain Monte Carlo procedure is described. We evaluate the stability of the sampling procedure and the predictive ability of the model compared against the current practice."

How'd I miss this when it came out?

"This paper develops a connection between the asymptotic stability of nonlinear filters and a notion of observability. We consider a general class of hidden Markov models in continuous time with compact signal state space, and call such a model observable if no two initial measures of the signal process give rise to the same law of the observation process. We demonstrate that observability implies stability of the filter, i.e., the filtered estimates become insensitive to the initial measure at large times. For the special case where the signal is a finite-state Markov process and the observations are of the white noise type, a complete (necessary and sufficient) characterization of filter stability is obtained in terms of a slightly weaker detectability condition. In addition to observability, the role of controllability in filter stability is explored. Finally, the results are partially extended to non-compact signal state spaces."