Multiple task transfer learning with small sample sizes

Prognosis, such as predicting mortality, is common in medicine. When confronted with small numbers of samples, as in rare medical conditions, the task is challenging. We propose a framework for classification with data with small numbers of samples. Conceptually, our solution is a hybrid of multi-task and transfer learning, employing data samples from source tasks as in transfer learning, but considering all tasks together as in multi-task learning. Each task is modelled jointly with other related tasks by directly augmenting the data from other tasks. The degree of augmentation depends on the task relatedness and is estimated directly from the data. We apply the model on three diverse real-world data sets (healthcare data, handwritten digit data and face data) and show that our method outperforms several state-of-the-art multi-task learning baselines. We extend the model for online multi-task learning where the model parameters are incrementally updated given new data or new tasks. The novelty of our method lies in offering a hybrid multi-task/transfer learning model to exploit sharing across tasks at the data-level and joint parameter learning.

A new transfer learning framework with application to model-agnostic multi-task learning

Learning from small number of examples is a challenging problem in machine learning. An effective way to improve the performance is through exploiting knowledge from other related tasks. Multi-task learning (MTL) is one such useful paradigm that aims to improve the performance through jointly modeling multiple related tasks. Although there exist numerous classification or regression models in machine learning literature, most of the MTL models are built around ridge or logistic regression. There exist some limited works, which propose multi-task extension of techniques such as support vector machine, Gaussian processes. However, all these MTL models are tied to specific classification or regression algorithms and there is no single MTL algorithm that can be used at a meta level for any given learning algorithm. Addressing this problem, we propose a generic, model-agnostic joint modeling framework that can take any classification or regression algorithm of a practitioner’s choice (standard or custom-built) and build its MTL variant. The key observation that drives our framework is that due to small number of examples, the estimates of task parameters are usually poor, and we show that this leads to an under-estimation of task relatedness between any two tasks with high probability. We derive an algorithm that brings the tasks closer to their true relatedness by improving the estimates of task parameters. This is achieved by appropriate sharing of data across tasks. We provide the detail theoretical underpinning of the algorithm. Through our experiments with both synthetic and real datasets, we demonstrate that the multi-task variants of several classifiers/regressors (logistic regression, support vector machine, K-nearest neighbor, Random Forest, ridge regression, support vector regression) convincingly outperform their single-task counterparts. We also show that the proposed model performs comparable or better than many state-of-the-art MTL and transfer learning baselines.

Flexible transfer learning framework for bayesian optimisation

Bayesian optimisation is an efficient technique to optimise functions that are expensive to compute. In this paper, we propose a novel framework to transfer knowledge from a completed source optimisation task to a new target task in order to overcome the cold start problem. We model source data as noisy observations of the target function. The level of noise is computed from the data in a Bayesian setting. This enables flexible knowledge transfer across tasks with differing relatedness, addressing a limitation of the existing methods. We evaluate on the task of tuning hyperparameters of two machine learning algorithms. Treating a fraction of the whole training data as source and the whole as the target task, we show that our method finds the best hyperparameters in the least amount of time compared to both the state-of-art and no transfer method.

From Concrete to Abstract: Teaching for Transfer of Learning when Using Manipulatives

One of the most important uses of manipulatives in a classroom is to aid a learner to make connection from tangible concrete object to its abstraction. In this paper we discuss how teacher educators can foster deeper understanding of how manipulatives facilitate student learning of math concepts by emphasizing the connection between concrete objects and math symbolization with, preservice elementary teachers, the future implementers of knowledge. We provide an example and a model, with specific steps of how teacher educators can effectively demonstrate connections between concrete objects and abstract math concepts.

Interocular transfer in perceptual learning of a pop-out discrimination task.

The specificity of the improvement in perceptual learning is often used to localize the neuronal changes underlying this type of adult plasticity. We investigated a visual texture discrimination task previously reported to be accomplished preattentively and for which learning-related changes were inferred to occur at a very early level of the visual processing stream. The stimulus was a matrix of lines from which a target popped out, due to an orientation difference between the three target lines and the background lines. The task was to report the global orientation of the target and was performed monocularly. The subjects' performance improved dramatically with training over the course of 2-3 weeks, after which we tested the specificity of the improvement for the eye trained. In all subjects tested, there was complete interocular transfer of the learning effect. The neuronal correlate of this learning are therefore most likely localized in a visual area where input from the two eyes has come together.

Multi-task transfer learning for in hospital-death prediction for ICU patients

Multi-Task Transfer Learning (MTTL) is an efficient approach for learning from inter-related tasks with small sample size and imbalanced class distribution. Since the intensive care unit (ICU) data set (publicly available in Physionet) has subjects from four different ICU types, we hypothesizethat there is an underlying relatedness amongst various ICU types. Therefore, this study aims to explore MTTL model for in-hospital mortality prediction of ICU patients. We used singletask learning (STL) approach on the augmented data as well as individual ICU data and compared the performance with the proposed MTTL model. As a performance measurement metrics, we used sensitivity (Sens), positive predictivity (+Pred), and Score. MTTL with class balancing showed the best performance with score of 0.78, 0.73, o.52 and 0.63 for ICU type 1(Coronary care unit), 2 (Cardiac surgery unit), 3 (Medical ICU) and 4 (Surgical ICU) respectively. In contrast the maximum score obtained using STL approach was 0.40 for ICU type 1 & 2. These results indicates that the performance of in-hospital mortality can be improved using ICU type information and by balancing the ’non-survivor’ class. The findings of the study may be useful for quantifying the quality of ICU care, managing ICU resources and selecting appropriate interventions.

Micropolar flow in a porous channel with high mass transfer

Two dimensional flow of a micropolar fluid in a porous channel is investigated. The flow is driven by suction or injection at the channel walls, and the micropolar model due to Eringen is used to describe the working fluid. An extension of Berman's similarity transform is used to reduce the governing equations to a set of non-linear coupled ordinary differential equations. The latter are solved for large mass transfer via a perturbation analysis where the inverse of the cross-flow Reynolds number is used as the perturbing parameter. Complementary numerical solutions for strong injection are also obtained using a quasilinearisation scheme, and good agreement is observed between the solutions obtained from the perturbation analysis and the computations.

The impact of unique characteristics of projects and project-based organisations on knowledge transfer

Knowledge has been recognised as an important organisational asset that increases in value when shared; the opposite to other organisational assets which decrease in value during their exploitation. Effective knowledge transfer in organisations helps to achieve and maintain competitive advantage and ultimately organisational success. So far, the research on knowledge transfer has focused on traditional (functional) organisations. Only recently has attention been directed towards knowledge transfer in projects. Existing research on project learning has recognised the need for knowledge transfer within and across projects in project-based organisations (PBOs). Most projects can provide valuable new knowledge from unexpected actions, approaches or problems experienced during the project phases. The aim of this paper is to demonstrate the impact of unique projects characteristics on knowledge transfer in PBO. This is accomplished through review of the literature and a series of interviews with senior project practitioners. The interviews complement the findings from the literature. Knowledge transfer in projects occurs by social communication and transfer of lessons learned where project management offices (PMOs) and project managers play significant roles in enhancing knowledge transfer and communication within the PBO and across projects. They act as connectors between projects and the PBO ‘hub’. Moreover, some project management processes naturally facilitate knowledge transfer across projects. On the other hand, PBOs face communication challenges due to unique and temporary characteristics of projects. The distance between projects and the lack or weakness of formal links across projects, create communication problems that impede knowledge transfer across projects. The main contribution of this paper is to demonstrate that both social communication and explicit informational channels play important role in inter-project knowledge transfer. Interviews also revealed the important role organisational culture play in knowledge transfer in PBOs.

Experience of problem-based learning (PBL) in virtual space : a phenomenographical study

This thesis reports the outcomes of an investigation into students’ experience of Problem-based learning (PBL) in virtual space. PBL is increasingly being used in many fields including engineering education. At the same time many engineering education providers are turning to online distance education. Unfortunately there is a dearth of research into what constitutes an effective learning experience for adult learners who undertake PBL instruction through online distance education. Research was therefore focussed on discovering the qualitatively different ways that students experience PBL in virtual space. Data was collected in an electronic environment from a course, which adopted the PBL strategy and was delivered entirely in virtual space. Students in this course were asked to respond to open-ended questions designed to elicit their learning experience in the course. Data was analysed using the phenomenographical approach. This interpretative research method concentrated on mapping the qualitative differences in students’ interpretations of their experience in the course. Five qualitatively different ways of experiencing were discovered: Conception 1: ‘A necessary evil for program progression’; Conception 2: ‘Developing skills to understand, evaluate, and solve technical Engineering and Surveying problems’; Conception 3: ‘Developing skills to work effectively in teams in virtual space’; Conception 4: ‘A unique approach to learning how to learn’; Conception 5: ‘Enhancing personal growth’. Each conception reveals variation in how students attend to learning by PBL in virtual space. Results indicate that the design of students’ online learning experience was responsible for making students aware of deeper ways of experiencing PBL in virtual space. Results also suggest that the quality and quantity of interaction with the team facilitator may have a significant impact on the student experience in virtual PBL courses. The outcomes imply pedagogical strategies can be devised for shifting students’ focus as they engage in the virtual PBL experience to effectively manage the student learning experience and thereby ensure that they gain maximum benefit. The results from this research hold important ramifications for graduates with respect to their ease of transition into professional work as well as their later professional competence in terms of problem solving, ability to transfer basic knowledge to real-life engineering scenarios, ability to adapt to changes and apply knowledge in unusual situations, ability to think critically and creatively, and a commitment to continuous life-long learning and self-improvement.

Developing capabilities for ongoing learning and change in intermediary programs

In dynamic environments, firms seek to build capabilities which will permit them to become innovation and change ready. Programs offered by intermediaries, while varying greatly in content and format, are designed to support those firms wishing to enhance their competitiveness. Firms which participate in intermediary programs have displayed their willingness to overcome deficiencies or barriers to competitiveness through acquiring knowledge which is external to the firm. This paper reports on interviews with 24 firms who were involved in a MAP or TAP program offered by QMI Solutions. The findings of the research suggest that knowledge intermediaries serve to disrupt organisational paths and in so doing establish mechanisms for ongoing learning and change. They do this first by disrupting the firm with a positive learning experience and also by establishing processes for developing new relationships and access to knowledge which are critical for learning and change. It is the experience of learning through knowledge exchange which can trigger the pursuit of new paths and it is the processes involving new relations and knowledge processing that provides the micro-foundations for ongoing learning and change. This suggests that the role of intermediaries goes well beyond merely knowledge transfer to include longer term effects on the capability of organisations to innovate, which is critical to economic competitiveness and the survival rate of firms.

Learning through collaboration and co-production in the Singapore film industry : opportunities for development

Co-production and strategic partnerships may generate valuable learning opportunities for firms to access to the knowledge and expertise of their partners. Such sharing and transfer of knowledge has become an increasingly common way for organising corporate finance and resources. However, not all collaborations result in a net positive experience for both partners. It can be a zero-sum game in which the partner learning the fastest dominates the relationship. In some cases, failure to gain access to partner knowledge results in unequal benefits accruing from such alliances. By examining the Singapore film industry from a learning perspective and taking into account particular forms of alliances, the study contributes to our understanding of the potential benefit and challenges of coproduction as a strategy for development.