Supervised latent dirichlet allocation models for efficient activity representation

Local spatio-temporal features with a Bag-of-visual words model is a popular approach used in human action recognition. Bag-of-features methods suffer from several challenges such as extracting appropriate appearance and motion features from videos, converting extracted features appropriate for classification and designing a suitable classification framework. In this paper we address the problem of efficiently representing the extracted features for classification to improve the overall performance. We introduce two generative supervised topic models, maximum entropy discrimination LDA (MedLDA) and class- specific simplex LDA (css-LDA), to encode the raw features suitable for discriminative SVM based classification. Unsupervised LDA models disconnect topic discovery from the classification task, hence yield poor results compared to the baseline Bag-of-words framework. On the other hand supervised LDA techniques learn the topic structure by considering the class labels and improve the recognition accuracy significantly. MedLDA maximizes likelihood and within class margins using max-margin techniques and yields a sparse highly discriminative topic structure; while in css-LDA separate class specific topics are learned instead of common set of topics across the entire dataset. In our representation first topics are learned and then each video is represented as a topic proportion vector, i.e. it can be comparable to a histogram of topics. Finally SVM classification is done on the learned topic proportion vector. We demonstrate the efficiency of the above two representation techniques through the experiments carried out in two popular datasets. Experimental results demonstrate significantly improved performance compared to the baseline Bag-of-features framework which uses kmeans to construct histogram of words from the feature vectors.

Autonomous vehicle path planning for persistence monitoring under uncertainty using Gaussian based Markov decision process

One of the main challenges facing online and offline path planners is the uncertainty in the magnitude and direction of the environmental energy because it is dynamic, changeable with time, and hard to forecast. This thesis develops an artificial intelligence for a mobile robot to learn from historical or forecasted data of environmental energy available in the area of interest which will help for a persistence monitoring under uncertainty using the developed algorithm.

Locally weighted learning methods for non-rigid robot control

This thesis develops a novel approach to robot control that learns to account for a robot's dynamic complexities while executing various control tasks using inspiration from biological sensorimotor control and machine learning. A robot that can learn its own control system can account for complex situations and adapt to changes in control conditions to maximise its performance and reliability in the real world. This research has developed two novel learning methods, with the aim of solving issues with learning control of non-rigid robots that incorporate additional dynamic complexities. The new learning control system was evaluated on a real three degree-of-freedom elastic joint robot arm with a number of experiments: initially validating the learning method and testing its ability to generalise to new tasks, then evaluating the system during a learning control task requiring continuous online model adaptation.

Learning the learning rate for prediction with expert advice

Most standard algorithms for prediction with expert advice depend on a parameter called the learning rate. This learning rate needs to be large enough to fit the data well, but small enough to prevent overfitting. For the exponential weights algorithm, a sequence of prior work has established theoretical guarantees for higher and higher data-dependent tunings of the learning rate, which allow for increasingly aggressive learning. But in practice such theoretical tunings often still perform worse (as measured by their regret) than ad hoc tuning with an even higher learning rate. To close the gap between theory and practice we introduce an approach to learn the learning rate. Up to a factor that is at most (poly)logarithmic in the number of experts and the inverse of the learning rate, our method performs as well as if we would know the empirically best learning rate from a large range that includes both conservative small values and values that are much higher than those for which formal guarantees were previously available. Our method employs a grid of learning rates, yet runs in linear time regardless of the size of the grid.

Superpixel-based appearance change prediction for long-term navigation across seasons

Changing environments pose a serious problem to current robotic systems aiming at long term operation under varying seasons or local weather conditions. This paper is built on our previous work where we propose to learn to predict the changes in an environment. Our key insight is that the occurring scene changes are in part systematic, repeatable and therefore predictable. The goal of our work is to support existing approaches to place recognition by learning how the visual appearance of an environment changes over time and by using this learned knowledge to predict its appearance under different environmental conditions. We describe the general idea of appearance change prediction (ACP) and investigate properties of our novel implementation based on vocabularies of superpixels (SP-ACP). Our previous work showed that the proposed approach significantly improves the performance of SeqSLAM and BRIEF-Gist for place recognition on a subset of the Nordland dataset under extremely different environmental conditions in summer and winter. This paper deepens the understanding of the proposed SP-ACP system and evaluates the influence of its parameters. We present the results of a large-scale experiment on the complete 10 h Nordland dataset and appearance change predictions between different combinations of seasons.

Curiosity driven reinforcement learning for motion planning on humanoids

Most previous work on artificial curiosity (AC) and intrinsic motivation focuses on basic concepts and theory. Experimental results are generally limited to toy scenarios, such as navigation in a simulated maze, or control of a simple mechanical system with one or two degrees of freedom. To study AC in a more realistic setting, we embody a curious agent in the complex iCub humanoid robot. Our novel reinforcement learning (RL) framework consists of a state-of-the-art, low-level, reactive control layer, which controls the iCub while respecting constraints, and a high-level curious agent, which explores the iCub's state-action space through information gain maximization, learning a world model from experience, controlling the actual iCub hardware in real-time. To the best of our knowledge, this is the first ever embodied, curious agent for real-time motion planning on a humanoid. We demonstrate that it can learn compact Markov models to represent large regions of the iCub's configuration space, and that the iCub explores intelligently, showing interest in its physical constraints as well as in objects it finds in its environment.

Career counseling : joint contributions of contextual action theory and the systems theory framework

The influence of constructivism and the ongoing drive for convergence, both of career theories and between theory and practice, have been key drivers in the career development literature for two decades (Patton, International Handbook of Career Guidance, 2008). Both contextual action theory and systems theory are derived from the root metaphor of contextualism, which has been proffered as a worldview to assist scientists and practitioners in organizing day-to-day experiential data. This chapter identifies the theoretical contributions of the Systems Theory Framework (STF) (Patton and McMahon, Career development and systems theory: A new development, 1999, Career psychology in South Africa, 2006) and Contextual Action Theory (Young and Valach, The future of career, 2000, Journal of Vocational Behavior 64:499–514, 2004; Young et al., Career choice and development, 1996, Career choice and development, 2002), each of which has advanced thinking in theory integration and in the integration between theory and practice in the career development and counseling field. Young et al. (Career development in childhood and adolescence, 2007) noted the connections between the Patton and McMahon systems theory approach and the contextual action theory approach and these connections will be highlighted in terms of the application of these theoretical developments to practice in career counseling, with a particular focus on the commonalities between the two approaches and what counselors can learn from each of them. In particular, this chapter will discuss common conceptual understandings and practice dimensions.

Stories have the power to save us: A neurological framework for the imperative to tell stories

The evolutionary advantage of humans is in our unique ability to process stories – we have highly evolved ‘narrative organs.’ Through storytelling, vicarious knowledge, even guarded knowledge, is used to help our species to survive. We learn, regardless of whether the story being told is ‘truth’ or ‘fiction.’ This article discusses how humans place themselves in stories, as both observer and participant, to create a ‘neural balance’ or sweet spot that allows them to be immersed in a story without being entirely threatened by it – and how this involvement in story is the formation of empathy – an empathy that is integral to forging a future humanity. It is through empathy, we argue, that stories have the power to save us.

Remaking Guangzhou: Political engagement and place-making on Sina Weibo

This study uses the concept of ‘place-making’ to consider political engagement on Sina Weibo, one of the most popular microblogging services in China. Besides articulating statepublic confrontation during major social controversies, Weibo has been used to recollect and renarrate the memories of a city, such as Guangzhou, where dramatic social and cultural changes took place during the economic reform era. The Chinese government’s ongoing project to create a culturally indifferent ‘national identity’ triggers a defensive response from local places. Through consuming news and information about leisure and entertainment in Guangzhou, the digital narration of the city becomes an important source for Guangzhou people to learn about their geo-identity, and the kind of rights and responsibility attaching to it.

Self-regulation support through music

A critical dimension of early learning competence in the year prior to school is self-regulation. Self-regulation enables children to manage their emotions and direct their attention, thinking, and actions to meet adaptive goals. These skills enhance young children's readiness to learn.

Co-creating stories in social learning systems : The role of community media and cultural organisations in disseminating knowledge

This paper investigates how community based media organisations are co-creative storytelling institutions, and how they learn to disseminate knowledge in a social learning system. Organisations involved in story co-creation are learning to create in fluid environments.They are project based, with a constant turnover of volunteers or staff. These organisations have to meet the needs of their funding bodies and their communities to remain sustainable. Learning is seen as dialogical, and this is also reflected in the nature of storytelling itself. These organisations must learn to meet the needs of their communities, who in turn learn from the organisation’s expertise in a facilitated setting. This learning is participatory and collaborative, and is often a mix of virtual and offline interaction. Such community-based organisations sit in the realm of a hybrid-learning environment; they are neither a formal educational institution like a college, nor do their volunteers produce outcomes in a professional capacity. Yet, they must maintain a certain level of quality outcomes from their contributors to be of continued value in their communities. Drawing from a larger research study, one particular example is that of the CitizenJ project. CitizenJ is hosted by a state cultural centre, and partnered with publishing partners in the community broadcasting sector. This paper explores how this project is a Community of Practice, and how it promotes ethical and best practice, meets contributors’ needs, emphasises the importance of facilitation in achieving quality outcomes, and the creation of projects for wider community and public interest.

Experiencing information use for early career academics’ learning: A knowledge ecosystem model

Purpose This study explores the informed learning experiences of early career academics while building their networks for professional and personal development. The notion that information and learning are inextricably linked via the concept of ‘informed learning’ is used as a conceptual framework to gain a clearer picture of what informs early career academics while they learn and how they experience using that which informs their learning within this complex practice: to build, maintain and utilise their developmental networks. Methodology This research employs a qualitative framework using a constructivist grounded theory approach (Charmaz, 2006). Through semi-structured interviews with a sample of fourteen early career academics from across two Australian universities, data were generated to investigate the research questions. The study used the methods of constant comparison to create codes and categories towards theme development. Further examination considered the relationship between thematic categories to construct an original theoretical model. Findings The model presented is a ‘knowledge ecosystem’, which represents the core informed learning experience. The model consists of informal learning interactions such as relating to information to create knowledge and engaging in mutually supportive relationships with a variety of knowledge resources found in people who assist in early career development. Originality/Value Findings from this study present an alternative interpretation of informed learning that is focused on processes manifesting as human interactions with informing entities revolving around the contexts of reciprocal human relationships.

Learning to lead: The development and testing of a model of leadership learning

Drawing upon an action learning perspective, we hypothesized that a leader’s learning of project leadership skills would be related to facilitative leadership, team reflexivity, and team performance. Secondly, we proposed that new and experienced leaders would differ in the amount they learn from their current and recent experience as project managers, and in the strength of the relationship between their self-reported learning, facilitative leadership, and team reflexivity. We conducted a 1-year longitudinal study of 50 R&D teams, led by 25 new and 25 experienced leaders, with 313 team members and 22 project customers, collecting both quantitative and qualitative data. We found evidence of a significant impact of the leader’s learning on subsequent facilitative leadership and team performance 8 and 12 months later, suggesting a lag between learning leadership skills and translating these skills into leadership behavior. The findings contribute to an understanding of how leaders consolidate their learned experience into facilitative leadership behavior.

Secret Millionares Club

Secret Millionaires Club is an animated series of 26 webisodes featuring Warren Buffett (CEO and largest shareholder of Berkshire Hathaway) as a secret mentor to a group of kids who learn practical life lessons during fun-filled adventures in business.

Online learning of autonomous helicopter control

This paper details the development of an online adaptive control system, designed to learn from the actions of an instructing pilot. Three learning architectures, single layer neural networks (SLNN), multi-layer neural networks (MLNN), and fuzzy associative memories (FAM) are considerd. Each method has been tested in simulation. While the SLNN and MLNN provided adequate control under some simulation conditions, the addition of pilot noise and pilot variation during simulation training caused these methods to fail.