We combined a spoken dialog system that we developed to deliver brief health interventions with the fully autonomous humanoid robot (NAO). The dialog system is based on a framework facilitating Markov decision processes (MDP). It is optimized using reinforcement learning (RL) algorithms with data we collected from real user interactions. The system begins to learn optimal dialog strategies for initiative selection and for the type of confirmations that it uses during theinteraction. The health intervention, delivered by a 3D character instead of the NAO, has already been evaluated, with positive results in terms of task completion, ease of use, and future intention to use the system.  The current spoken dialog system for the humanoid robot is a novelty and exists so far as a proof ofconcept.

Task starvation leads to huge variation in the completion times of the tasks posted on to the crowd. The price offered to a given task together with the dynamics of the crowd at the time of posting affect its completion time. Large organizations/requesters who frequent the crowd at regular intervals in order to get their tasks done desire predictability in completion times of the tasks. Thus, such requesters have to take into account the crowd dynamics at the time of posting the tasks and price them accordingly. In this work, we study an instance of the pricing problem and propose a solution based on the framework of Markov Decision Processes (MDPs).

While temporal behavioral patterns can be discerned to underlie real crowd work, prior studies have typically modeled worker performance under a simplified i.i.d. assumption. To better model such temporal worker behavior, we propose a time-series label prediction model for crowd work. This latent variable model captures and summarizes past worker behavior, enabling us to better predict the quality of each worker's next label. Given inherent uncertainty in prediction, we also investigate a decision reject option to balance the tradeoff between prediction accuracy vs. coverage. Results show our model improves accuracy of both label prediction on real crowd worker data, as well as data quality overall.

An ideal crowdsourcing or citizen-science system would route tasks to the most appropriate workers, but the best assignment is unclear because workers have varying skill, tasks have varying difficulty, and assigning several workers to a single task may significantly improve output quality. This paper defines a space of task routing problems, proves that even the simplest is NP-hard, and develops several approximation algorithms for parallel routing problems. We show that an intuitive class of requesters' utility functions is submodular, which lets us provide iterative methods for dynamically allocating batches of tasks that make near-optimal use of available workers in each round. Experiments with live oDesk workers show that our task routing algorithm uses only 48% of the human labor compared to the commonly used round-robin strategy. Further, we provide versions of our task routing algorithm which enable it to scale to large numbers of workers and questions and to handle workers with variable response times while still providing significant benefit over common baselines.

We consider the problem of learning to optimize an unknown Markov decision process (MDP). We show that, if the MDP can be parameterized within some known function class, we can obtain regret bounds that scale with the dimensionality, rather than cardinality, of the system. We characterize this dependence explicitly as $\tilde{O}(\sqrt{d_K d_E T})$ where $T$ is time elapsed, $d_K$ is the Kolmogorov dimension and $d_E$ is the \emph{eluder dimension}. These represent the first unified regret bounds for model-based reinforcement learning and provide state of the art guarantees in several important settings. Moreover, we present a simple and computationally efficient algorithm \emph{posterior sampling for reinforcement learning} (PSRL) that satisfies these bounds.

We demonstrate that there is significant redundancy in the parameterization of several deep learning models. Given only a few weight values for each feature it is possible to accurately predict the remaining values. Moreover, we show that not only can the parameter values be predicted, but many of them need not be learned at all. We train several different architectures by learning only a small number of weights and predicting the rest. In the best case we are able to predict more than 95% of the weights of a network without any drop in accuracy.

Restricted Boltzmann machines (RBMs) are powerful machine learning models, but learning and some kinds of inference in the model require sampling-based approximations, which, in classical digital computers, are implemented using expensive MCMC. Physical computation offers the opportunity to reduce the cost of sampling by building physical systems whose natural dynamics correspond to drawing samples from the desired RBM distribution. Such a system avoids the burn-in and mixing cost of a Markov chain. However, hardware implementations of this variety usually entail limitations such as low-precision and limited range of the parameters and restrictions on the size and topology of the RBM. We conduct software simulations to determine how harmful each of these restrictions is. Our simulations are based on the D-Wave Two computer, but the issues we investigate arise in most forms of physical computation. Our findings suggest that designers of new physical computing hardware and algorithms for physical computers should focus their efforts on overcoming the limitations imposed by the topology restrictions of currently existing physical computers.

Runge-Kutta methods are the classic family of solvers for ordinary differential equations (ODEs), and the basis for the state of the art. Like most numerical methods, they return point estimates. We construct a family of probabilistic numerical methods that instead return a Gauss-Markov process defining a probability distribution over the ODE solution. In contrast to prior work, we construct this family such that posterior means match the outputs of the Runge-Kutta family exactly, thus inheriting their proven good properties. Remaining degrees of freedom not identified by the match to Runge-Kutta are chosen such that the posterior probability measure fits the observed structure of the ODE. Our results shed light on the structure of Runge-Kutta solvers from a new direction, provide a richer, probabilistic output, have low computational cost, and raise new research questions.

Although EEG signals due to their high temporal resolution show highly stochastic temporal evolution, it has been found that the scalp potential topographies are not so random and follow finite sets of small number of quasi-stable patterns which are termed as microstates [2]. Recently, Jamal et al. [3] investigated the temporal evolution of the frequency band-specific phase difference topographies to find periods of phase locking in multichannel EEG signals. It has been found in [4] that the phase difference topographies do not change abruptly and microstate-like quasi-stable phase locked patterns are observed in a temporal resolution of the order of milliseconds. These small number of stable phase synchronized patterns are termed as synchrostates, which switches from one to the other within the time interval of a cognitive task. The existence of synchrostates during face perception tasks was first observed in the beta (β) band (13-30 Hz) with different ensembles of EEG signals [4]. For similar visual stimuli, the interstate switching patterns only slightly change among different ensembles or trials [4], however it is different for different stimuli and also across different groups of people [3].

Mass spectrometry can involve two soft ionization techniques: matrix-assisted laser desorption ionization (MALDI) and surface-enhanced laser desorption and ionization (SELDI). For each analyzed fluid sample, MALDI or SELDI hardwares generate a high-dimensional mass spectrum, recording between 10,000 and 20,000 "mass-to-charge (m/z) ratios" corresponding to the ionized peptides present in the fluid sample, as well as "intensities" roughly quantifying the concentrations of these peptides in the sample. Generally m/z ratios take values anywhere between 200 and 20,000 Daltons, and are acquired with a known relative accuracy ρ which depends on the acquisition modalities, and ranges from 0.1% to 0.3%. Analyzing this type of high dimensional data oftern requires specialized software tools, implementing sophisticated machine learning techniques such as SVM (support vector machines) (Li and others (2004), Yu and others (2005)), artificial neural networks (Ball and others (2002)), or random forests (Izmirlian (2004)). These techniques typically generate "black-box" classifiers, which often reach good discrimination levels between cancerous and control groups, but are difficult to interpret biologically in terms of characteristic biomarkers patterns. This often leads to unexpected performance variations on totally new data sets. To develop clinically usable software tools for analysis of mass spectra acuired by MALDI or SELDI hardwares, a key step is to implement automated discovery of explicit "signatures", i.e. short lists of proteomic biomarkers with high discriminating powers between cancer groups (Yasui and others (2003)). Some easily interpretable automatic classifiers, such as linear combinations of biomarker weights (Wang and Chang (2011)), can be found in previous studies, but these approaches do not attempt to quantify the discriminating impact of simultaneous presence for specific pairs of biomarkers.