Resource constraints frequently complicate multi-agent planning problems. Existing algorithms for resource-constrained, multi-agent planning problems rely on the assumption that the constraints are deterministic. However, frequently resource constraints are themselves subject to uncertainty from external influences. Uncertainty about constraints is especially challenging when agents must execute in an environment where communication is unreliable, making on-line coordination difficult. In those cases, it is a significant challenge to find coordinated allocations at plan time depending on availability at run time. To address these limitations, we propose to extend algorithms for constrained multi-agent planning problems to handle stochastic resource constraints. We show how to factorize resource limit uncertainty and use this to develop novel algorithms to plan policies for stochastic constraints. We evaluate the algorithms on a search-and-rescue problem and on a power-constrained planning domain where the resource constraints are decided by nature. We show that plans taking into account all potential realizations of the constraint obtain significantly better utility than planning for the expectation, while causing fewer constraint violations.

In the context of fair allocation of indivisible items, fairness concepts often compare the satisfaction of an agent to the satisfaction she would have from items that are not allocated to her: in particular, envy-freeness requires that no agent prefers the share of someone else to her own share. We argue that these notions could also be defined relative to the knowledge that an agent has on how the items that she does not receive are distributed among other agents. We define a family of epistemic notions of envy-freeness, parameterized by a social graph, where an agent observes the share of her neighbours but not of her non-neighbours. We also define an intermediate notion between envy-freeness and proportionality, also parameterized by a social graph. These weaker notions of envy-freeness are useful when seeking a fair allocation, since envy-freeness is often too strong. We position these notions with respect to known ones, thus revealing new rich hierarchies of fairness concepts. Finally, we present a very general framework that covers all the existing and many new fairness concepts.

Signal processing based filtering methods. Those approaches Data sets collected from a wide variety of research disciplines, implicitly assume that the "normal" component including computer science, economic, biology and of the time series has a sparse representation in the frequency social science, are in the form of multiple co-evolving time or wavelet domain. Hence the outlier detection problem is reduced series. In this work, we consider the task of outlier (or novelty) to a spectral analysis using low pass or band pass filters, detection given the aforementioned data type. The core or is solved by denoising/signal reconstruction using spectral difficulty, however, is to integrate both the temporal dependence or wavelet techniques (Mallat 2008). It is worth pointing out and the interactions among correlated time series for that the signal-processing-based methods have close ties with overall modeling and learning.

We tackle the problem of identifiability and efficient learning of mixtures of Random Utility Models (RUMs). We show that when the PDFs of utility distributions are symmetric, the mixture of k RUMs (denoted by k-RUM) is not identifiable when the number of alternatives m is no more than 2k-1. On the other hand, when m ≥ max{4k-2,6}, any k-RUM is generically identifiable. We then propose three algorithms for learning mixtures of RUMs: an EM-based algorithm, which we call E-GMM, a direct generalized-method-of-moments (GMM) algorithm, and a sandwich (GMM-E-GMM) algorithm that combines the other two. Experiments on synthetic data show that the sandwich algorithm achieves the highest statistical efficiency and GMM is the most computationally efficient. Experiments on real-world data at Preflib show that Gaussian k-RUMs provide better fitness than a single Gaussian RUM, the Plackett-Luce model, and mixtures of Plackett-Luce models w.r.t. commonly-used model fitness criteria. To the best of our knowledge, this is the first work on learning mixtures of general RUMs.

Cross-domain data reconstruction methods derive a shared transformation across source and target domains. These methods usually make a specific assumption on noise, which exhibits limited ability when the target data are contaminated by different kinds of complex noise in practice. To enhance the robustness of domain adaptation under severe noise conditions, this paper proposes a novel reconstruction based algorithm in an information-theoretic setting. Specifically, benefiting from the theoretical property of correntropy, the proposed algorithm is distinguished with: detecting the contaminated target samples without making any specific assumption on noise; greatly suppressing the negative influence of noise on cross-domain transformation. Moreover, a relative entropy based regularization of the transformation is incorporated to avoid trivial solutions with the reaped theoretic advantages, i.e., non-negativity and scale-invariance. For optimization, a half-quadratic technique is developed to minimize the non-convex information-theoretic objectives with explicitly guaranteed convergence. Experiments on two real-world domain adaptation tasks demonstrate the superiority of our method.

We propose a Las Vegas transformation of Markov Chain Monte Carlo (MCMC) estimators of Restricted Boltzmann Machines (RBMs). We denote our approach Markov Chain Las Vegas (MCLV). MCLV gives statistical guarantees in exchange for random running times. MCLV uses a stopping set built from the training data and has maximum number of Markov chain steps K (referred as MCLV-K). We present a MCLV-K gradient estimator (LVS-K) for RBMs and explore the correspondence and differences between LVS-K and Contrastive Divergence (CD-K), with LVS-K significantly outperforming CD-K training RBMs over the MNIST dataset, indicating MCLV to be a promising direction in learning generative models.

Feature selection is effective in preparing high-dimensional data for a variety of learning tasks such as classification, clustering and anomaly detection. A vast majority of existing feature selection methods assume that all instances share some common patterns manifested in a subset of shared features. However, this assumption is not necessarily true in many domains where data instances could show high individuality. For example, in the medical domain, we need to capture the heterogeneous nature of patients for personalized predictive modeling, which could be characterized by a subset of instance-specific features. Motivated by this, we propose to study a novel problem of personalized feature selection. In particular, we investigate the problem in an unsupervised scenario as label information is usually hard to obtain in practice. To be specific, we present a novel unsupervised personalized feature selection framework UPFS to find some shared features by all instances and instance-specific features tailored to each instance. We formulate the problem into a principled optimization framework and provide an effective algorithm to solve it. Experimental results on real-world datasets verify the effectiveness of the proposed UPFS framework.

Dialogue Act recognition associate dialogue acts (i.e., semantic labels) to utterances in a conversation. The problem of associating semantic labels to utterances can be treated as a sequence labeling problem. In this work, we build a hierarchical recurrent neural network using bidirectional LSTM as a base unit and the conditional random field (CRF) as the top layer to classify each utterance into its corresponding dialogue act. The hierarchical network learns representations at multiple levels, i.e., word level, utterance level, and conversation level. The conversation level representations are input to the CRF layer, which takes into account not only all previous utterances but also their dialogue acts, thus modeling the dependency among both, labels and utterances, an important consideration of natural dialogue. We validate our approach on two different benchmark data sets, Switchboard and Meeting Recorder Dialogue Act, and show performance improvement over the state-of-the-art methods by 2.2% and 4.1% absolute points, respectively. It is worth noting that the inter-annotator agreement on Switchboard data set is 84%, and our method is able to achieve the accuracy of about 79% despite being trained on the noisy data.

Not like numerical data clustering, nominal data clustering is a very difficult problem because there exists no natural relative ordering between nominal attribute values. This paper mainly aims to make the Euclidean distance measure appropriate to nominal data clustering, and the core idea is the attribute value embedding, namely, transforming each nominal attribute value into a numerical vector. This embedding method consists of four steps. In the first step, the weights, which can quantify the amount of information in attribute values, is calculated for each value in each nominal attribute based on each object and its k nearest neighbors. In the second step, an intra-attribute value similarity matrix is created for each nominal attribute by using the attribute value's weights. In the third step, for each nominal attribute, we find another attribute with the maximal dependence on it, and build an inter-attribute value similarity matrix on the basis of the attribute value's weights related to these two attributes. In the last step, a diffusion matrix of each nominal attribute is constructed by the tensor product graph diffusion process, and this step can cause the acquired value embedding to contain simultaneously the intra- and inter-attribute value similarities information. To evaluate the effectiveness of our proposed method, experiments are done on 10 data sets. Experimental results demonstrate that our method not only enables the Euclidean distance to be used for nominal data clustering, but also can acquire the better clustering performance than several existing state-of-the-art approaches.

Predictive state representations (PSR) have emerged as a powerful method for modelling partially observable environments. PSR learning algorithms can build models for predicting all observable variables, or predicting only some of them conditioned on others (e.g., actions or exogenous variables). In the latter case, which we call conditional modelling, the accuracy of different estimates of the conditional probabilities for a fixed dataset can vary significantly, due to the limited sampling of certain conditions. This can have negative consequences on the PSR parameter estimation process, which are not taken into account by the current state-of-the-art PSR spectral learning algorithms. In this paper, we examine closely conditional modelling within the PSR framework. We first establish a new positive but surprisingly non-trivial result: a conditional model can never be larger than the complete model. Then, we address the core shortcoming of existing PSR spectral learning methods for conditional models by incorporating an additional step in the process, which can be seen as a type of matrix denoising. We further refine this objective by adding penalty terms for violations of the system dynamics matrix structure, which improves the PSR predictive performance. Empirical evaluations on both synthetic and real datasets highlight the advantages of the proposed approach.