Most approaches to causal discovery assume a fixed (or time-invariant) causal model; however, in practical situations, especially in neuroscience and economics, causal relations might be time-dependent for various reasons. This paper aims to identify the time-dependent causal relations from observational data. We consider general formulations for time-varying causal modeling on stochastic processes, which can also capture the causal influence from a certain type of unobserved confounders.  We focus on two issues: one is whether such a causal model, including the causal direction, is identifiable from observational data; the other is how to estimate such a model in a principled way. We show that under appropriate assumptions, the causal structure is identifiable according to our formulated model. We then propose a principled way for its estimation by extending Gaussian Process regression, which enables an automatic way to learn how the causal model changes over time. Experimental results on both artificial and real data demonstrate the practical usefulness of time-dependent causal modeling and the effectiveness of the proposed approach for estimation.

Low rank matrix recovery has shown its importance as a theoretic foundation in many areas of information processing. Its solutions are usually obtained in batch mode that requires to load all the data into memory during processing, and thus are hardly applicable on large scale data. Moreover, a fraction of data may be severely contaminated by outliers, which makes accurate recovery significantly more challenging. This paper proposes a novel online robust low rank matrix recovery method to address these difficulties. In particular, we first introduce an online algorithm to solve the problem of low rank matrix completion. Then we move on to low rank matrix recovery from observations with intensive outliers. The outlier support is robustly estimated from a perspective of mixture model. Experiments on both synthetic and real data are conducted to demonstrate the efficacy of our method and show its superior performance over the state-of-the-arts.

Model selection is an important problem of cost-sensitive SVM (CS-SVM). Although using solution path to find global optimal parameters is a powerful method for model selection, it is a challenge to extend the framework to solve two regularization parameters of CS-SVM simultaneously. To overcome this challenge, we make three main steps in this paper. (i) A critical-regions-based bi-parameter space partition algorithm is proposed to present all piecewise linearities of CS-SVM. (ii) An invariant-regions-based bi-parameter space partition algorithm is further proposed to compute empirical errors for all parameter pairs. (iii) The global optimal solutions for K-fold cross validation are computed by superposing K invariant region based bi-parameter space partitions into one. The three steps constitute the model selection of CS-SVM which can find global optimal parameter pairs in K-fold cross validation. Experimental results on seven normal datsets and four imbalanced datasets, show that our proposed method has better generalization ability and than various kinds of grid search methods, however, with less running time.

In this paper we investigate the use of coactive learning in a multitask setting. In coactive learning, an expert presents the learner with a problem and the learner returns a candidate solution. The expert then improves on the solution if necessary and presents the improved solution to the learner. The goal for the learner is to learn to produce solutions which cannot be further improved by the expert while minimizing the average expert effort. In this paper, we consider the setting where there are multiple experts (tasks), and in each iteration one expert presents a problem to the learner. While the experts are expected to have different solution preferences, they are also assumed to share similarities, which should enable generalization across experts. We analyze several algorithms for this setting and derive bounds on the average expert effort during learning. Our main contribution is the balanced Perceptron algorithm, which is the first coactive learning algorithm that is both able to generalize across experts when possible, while also guaranteeing convergence to optimal solutions for individual experts. Our experiments in three domains confirm that this algorithm is effective in the multitask setting, compared to natural baselines.

This paper studies an interesting problem: is it possible to predict the crowd opinion about a movie before the movie is actually released? The crowd opinion is here expressed by the distribution of ratings given by a sufficient amount of people. Consequently, the pre-release crowd opinion prediction can be regarded as a Label Distribution Learning (LDL) problem. In order to solve this problem, a Label Distribution Support Vector Regressor (LDSVR) is proposed in this paper. The basic idea of LDSVR is to fit a sigmoid function to each component of the label distribution simultaneously by a multi-output support vector machine. Experimental results show that LDSVR can accurately predict peoples’s rating distribution about a movie just based on the pre-release metadata of the movie.

Hierarchical Reinforcement Learning (HRL) outperforms many ‘flat’ Reinforcement Learning (RL) algorithms in some application domains. However, HRL may need longer time to obtain the optimal policy because of its large action space. Potential Based Reward Shaping (PBRS) has been widely used to incorporate heuristics into flat RL algorithms so as to reduce their exploration. In this paper, we investigate the integration of PBRS and HRL, and propose a new algorithm: PBRS-MAXQ-0. We prove that under certain conditions, PBRS-MAXQ-0 is guaranteed to converge. Empirical results show that PBRS-MAXQ-0 significantly outperforms MAXQ-0 given good heuristics, and can converge even when given misleading heuristics.

Most multi-label domains lack an authoritative taxonomy. Therefore, different taxonomies are commonly used in the same domain, which results in complications. Although this situation occurs frequently, there has been little study of it using a principled statistical approach. Given that (1) different taxonomies used in the same domain are generally founded on the same latent semantic space, where each possible label set in a taxonomy denotes a single semantic concept, and that (2) crowdsourcing is beneficial in identifying relationships between semantic concepts and instances at low cost, we proposed a novel probabilistic cascaded method for establishing a semantic matching function in a crowdsourcing setting that maps label sets in one (source) taxonomy to label sets in another (target) taxonomy in terms of the semantic distances between them. The established function can be used to detect the associated label set in the target taxonomy for an instance directly from its associated label set in the source taxonomy without any extra effort. Experimental results on real-world data (emotion annotations for narrative sentences) demonstrated that the proposed method can robustly establish semantic matching functions exhibiting satisfactory performance from a limited number of crowdsourced annotations.

Solving multi-manifolds clustering problems that include delineating and resolving multiple intersections is a very challenging problem. In this paper we propose a novel procedure for clustering intersecting multi-manifolds and delineating junctions in high dimensional spaces. We propose to explicitly and directly resolve ambiguities near the intersections by using 2 properties: One is the position of the data points in the vicinity of the detected intersection; the other is the reliable estimation of the tangent spaces away from the intersections. We experiment with our method on a wide range of geometrically complex settings of convoluted intersecting manifolds, on which we demon- strate higher clustering performance than the state of the art. This includes tackling challenging geometric structures such as when the tangent spaces at the intersections points are not orthogonal.

Most logic-based machine learning algorithms rely on an Occamist bias where textual complexity of hypotheses is minimised. Within Inductive Logic Programming (ILP), this approach fails to distinguish between the efficiencies of hypothesised programs, such as quick sort (O(n log n)) and bubble sort (O(n 2 )). This paper addresses this issue by considering techniques to minimise both the textual complexity and resource complexity of hypothesised robot strategies. We develop a general framework for the problem of minimising resource complexity and show that on two robot strategy problems, 1) Postman 2) Sorter (recursively sort letters for delivery), the theoretical resource complexities of optimal strategies vary depending on whether objects can be composed within a strategy. The approach considered is an extension of Meta-Interpretive Learning (MIL), a recently developed paradigm in ILP which supports predicate invention and the learning of recursive logic programs. We introduce a new MIL implementation, Metagol O , and prove its convergence, with increasing numbers of randomly chosen examples to optimal strategies of this kind. Our experiments show that Metagol O learns theoretically optimal robot sorting strategies, which is in agreement with the theoretical predictions showing a clear divergence in resource requirements as the number of objects grows. To the authors’ knowledge this paper is the first demonstration of a learning algorithm able to learn optimal resource complexity robot strategies and algorithms for sorting lists.

In repeated stochastic games (RSGs), an agent must quickly adapt to the behavior of previously unknown associates, who may themselves be learning. This machine-learning problem is particularly challenging due, in part, to the presence of multiple (even infinite) equilibria and inherently large strategy spaces. In this paper, we introduce a method to reduce the strategy space of two-player general-sum RSGs to a handful of expert strategies. This process, called mega, effectually reduces an RSG to a bandit problem. We show that the resulting strategy space preserves several important properties of the original RSG, thus enabling a learner to produce robust strategies within a reasonably small number of interactions. To better establish strengths and weaknesses of this approach, we empirically evaluate the resulting learning system against other algorithms in three different RSGs.