We develop a new model and algorithms for machine learning-based learning analytics, which estimate a learner's knowledge of the concepts underlying a domain, and content analytics, which estimate the relationships among a collection of questions and those concepts. Our model represents the probability that a learner provides the correct response to a question in terms of three factors: their understanding of a set of underlying concepts, the concepts involved in each question, and each question's intrinsic difficulty. We estimate these factors given the graded responses to a collection of questions. The underlying estimation problem is ill-posed in general, especially when only a subset of the questions are answered. The key observation that enables a well-posed solution is the fact that typical educational domains of interest involve only a small number of key concepts. Leveraging this observation, we develop both a bi-convex maximum-likelihood and a Bayesian solution to the resulting SPARse Factor Analysis (SPARFA) problem. We also incorporate user-defined tags on questions to facilitate the interpretability of the estimated factors. Experiments with synthetic and real-world data demonstrate the efficacy of our approach. Finally, we make a connection between SPARFA and noisy, binary-valued (1-bit) dictionary learning that is of independent interest.

In some reinforcement learning problems an agent may be provided with a set of input policies, perhaps learned from prior experience or provided by advisors. We present a reinforcement learning with policy advice (RLPA) algorithm which leverages this input set and learns to use the best policy in the set for the reinforcement learning task at hand. We prove that RLPA has a sub-linear regret of \tilde O(\sqrt{T}) relative to the best input policy, and that both this regret and its computational complexity are independent of the size of the state and action space. Our empirical simulations support our theoretical analysis. This suggests RLPA may offer significant advantages in large domains where some prior good policies are provided.

This paper establishes theoretical bonafides for implicit concurrent multivariate effect evaluation--implicit concurrency for short---a broad and versatile computational learning efficiency thought to underlie general-purpose, non-local, noise-tolerant optimization in genetic algorithms with uniform crossover (UGAs). We demonstrate that implicit concurrency is indeed a form of efficient learning by showing that it can be used to obtain close-to-optimal bounds on the time and queries required to approximately correctly solve a constrained version (k=7, \eta=1/5) of a recognizable computational learning problem: learning parities with noisy membership queries. We argue that a UGA that treats the noisy membership query oracle as a fitness function can be straightforwardly used to approximately correctly learn the essential attributes in O(log^1.585 n) queries and O(n log^1.585 n) time, where n is the total number of attributes. Our proof relies on an accessible symmetry argument and the use of statistical hypothesis testing to reject a global null hypothesis at the 10^-100 level of significance. It is, to the best of our knowledge, the first relatively rigorous identification of efficient computational learning in an evolutionary algorithm on a non-trivial learning problem.

We explore the effect of introducing prior information into the intermediate level of neural networks for a learning task on which all the state-of-the-art machine learning algorithms tested failed to learn. We motivate our work from the hypothesis that humans learn such intermediate concepts from other individuals via a form of supervision or guidance using a curriculum. The experiments we have conducted provide positive evidence in favor of this hypothesis. In our experiments, a two-tiered MLP architecture is trained on a dataset with 64x64 binary inputs images, each image with three sprites. The final task is to decide whether all the sprites are the same or one of them is different. Sprites are pentomino tetris shapes and they are placed in an image with different locations using scaling and rotation transformations. The first part of the two-tiered MLP is pre-trained with intermediate-level targets being the presence of sprites at each location, while the second part takes the output of the first part as input and predicts the final task's target binary event. The two-tiered MLP architecture, with a few tens of thousand examples, was able to learn the task perfectly, whereas all other algorithms (include unsupervised pre-training, but also traditional algorithms like SVMs, decision trees and boosting) all perform no better than chance. We hypothesize that the optimization difficulty involved when the intermediate pre-training is not performed is due to the {\em composition} of two highly non-linear tasks. Our findings are also consistent with hypotheses on cultural learning inspired by the observations of optimization problems with deep learning, presumably because of effective local minima.

This paper presents a novel framework for online incremental place recognition in an indoor environment. The framework addresses the scenario in which scene images are gradually obtained during long-term operation in the real-world indoor environment. Multiple users may interact with the classification system and confirm either current or past prediction results; the system then immediately updates itself to improve the classification system. This framework is based on the proposed \emph{n}-value self-organizing and incremental neural network (\emph{n}-SOINN), which has been derived by modifying the original SOINN to be appropriate for use in scene recognition. The evaluation was performed on the standard MIT 67-category indoor scene dataset and shows that the proposed framework achieves the same accuracy as that of the state-of-the-art offline method, while the computation time of the proposed framework is significantly faster and fully incremental update is allowed. Additionally, a small extra set of training samples is incrementally given to the system to simulate the incremental learning situation. The result shows that the proposed framework can leverage such additional samples and achieve the state-of-the-art result.

Given a classification task, what is the best way to teach the resulting boundary to a human? While machine learning techniques can provide excellent methods for finding the boundary, including the selection of examples in an online setting, they tell us little about how we would teach a human the same task. We propose to investigate the problem of example selection and presentation in the context of teaching humans, and explore a variety of mechanisms in the interests of finding what may work best. In particular, we begin with the baseline of random presentation and then examine combinations of several mechanisms: the indication of an example’s relative difficulty, the use of the shaping heuristic from the cognitive science literature (moving from easier examples to harder ones), and a novel kernel-based “coverage model” of the subject’s mastery of the task. From our experiments on 54 human subjects learning and performing a pair of synthetic classification tasks via our teaching system, we found that we can achieve the greatest gains with a combination of shaping and the coverage model.

In spatio-temporal reasoning, granularity is one of the factors to be considered when aiming at an effective and efficient representation of space and time. There is a large body of work which addresses the issue of granularity by representing space and time on a qualitative level. Other approaches use a predefined scale which implicitly determines granularity (e.g., seconds, minutes, hours, days, month, etc.). However, there are situations where the right level of granularity is unknown in the beginning, and is only determined in the problem solving process itself. This is the case in machine learning, where the learner has to find a representation for a problem and with that the right granularity for representing space and time. This paper introduces an algorithm which determines the most appropriate level of granularity during training. It uses several description logic learners as the learners, and the positive and negative examples presented to them as the determinators for refining coarse temporal representations to the most appropriate level of granularity.

In exploratory domains, agents' actions map onto logs of behavior that include switching between activities, extraneous actions, and mistakes. These aspects create a challenging plan recognition problem. This paper presents a new algorithm for inferring students' activities in exploratory domains that is evaluated empirically using a new type of flexible and open-ended educational software for science education. Such software has been shown to provide a rich educational environment for students, but challenge teachers to keep track of students' progress and to assess their performance. The algorithm decomposes students’ complete interaction histories to create hierarchies of interdependent tasks that describe their activities using the software. It matches students' actions to a predefined grammar in a way that reflects that students solve problems in a modular fashion but may still interleave between their activities. The algorithm was empirically evaluated on people’s interaction with two separate software systems for simulating a chemistry laboratory and for statistics education. It was separately compared to the state-of-the-art recognition algorithms for each of the software. The results show that the algorithm was able to correctly infer students' activities significantly more often than the state-of-the-art, and was able to generalize to both of the software systems with no intervention.

One of the major challenges to building intelligent educational software is determining what kinds of feedback to give learners. Useful feedback makes use of models of domain-specific knowledge, especially models that are commonly held by potential students. To empirically determine what these models are, student data can be clustered to reveal common misconceptions or common problem-solving strategies. This paper describes how analogical retrieval and generalization can be used to cluster automatically analyzed hand-drawn sketches incorporating both spatial and conceptual information. We use this approach to cluster a corpus of hand-drawn student sketches to discover common answers. Common answer clusters can be used for the design of targeted feedback and for assessment.

We describe a constraint-based timetabling system that was developed for the dental school based at Cork University Hospital in Ireland.This system has been deployed since 2010.Dental school timetabling differs from other university course scheduling in that certain clinic sessions can be used by multiple courses at the same time, provided a limit on room capacity is satisfied.Starting from a constraint programming solution using a web interface, we have moved to a mixed integer programming-based solver to deal with multiple objective functions, along with a dedicated Java application, which provides a rich user interface.Solutions for the years 2010, 2011 and 2012 have been used in the dental school, replacing a manual timetabling process, which could no longer cope with increasing student numbers and resulting resource bottlenecks.The use of the automated system allowed the dental school to increase student numbers to the maximum possible given the available resources.It also provides the school with a valuable "what-if" analysis tool.