Human Learning Systems Research Articles

Causal representation learning through higher-level information extraction

The large gap between the generalization level of state-of-the-art machine learning and human learning systems calls for the development of artificial intelligence (AI) models that are truly inspired by human cognition. In tasks related to image analysis, searching for pixel-level regularities has reached a power of information extraction still far from what humans capture with image-based observations. This leads to poor generalization when even small shifts occur at the level of the observations. We explore a perspective on this problem that is directed to learning the generative process with causality-related foundations, using models capable of combining symbolic manipulation, probabilistic reasoning and pattern recognition abilities. We briefly review and explore connections of research from machine learning, cognitive science and related fields of human behavior to support our perspective for the direction to more robust and human-like artificial learning systems.

Developing a Human Learning Systems approach for inter‐agency collaboration: Lessons from the UK NHS

Developing a Human Learning Systems approach for inter‐agency collaboration: Lessons from the UK NHS

Indoor air quality prediction modeling for a naturally ventilated fitness building using RNN-LSTM artificial neural networks

Indoor air quality prediction modeling for a naturally ventilated fitness building using RNN-LSTM artificial neural networks

Chronic pain, complexity and a suggested role for the osteopathic profession

Chronic pain is common and has a substantial impact on individuals’ lives and the society in which they live. However, current strategies for managing chronic pain have low to moderate efficacy. We present a complexity informed model of managing chronic pain.The authors propose that the biological, psychological and social factors influencing the experience of chronic pain can be framed as elements of a complex system (when ‘system’ is understood as a mental model of a set of relationships in the world) and discuss the advantages of this approach.The ‘Learning Cycles’ framework of Human Learning Systems is suggested as an appropriate approach to managing chronic pain. There are five elements of the Learning Cycle; ‘understanding the system’, ‘co-design’, ‘experimentation exploration’, ‘embedding and influencing’ and ‘system stewardship’. The authors describe how this could be implemented in osteopathic practice.The implications of a complexity approach with respect to evidence-informed as opposed to evidence-based practice is briefly discussed. Further, the opportunity for osteopathy to be at the vanguard of broader movement for change in public service and healthcare is highlighted.Applying Learning Cycles to the context of managing chronic pain requires further research to explore its utility. Learning Cycles provide an action-research model that would supply the rich data by which to understand the efficacy of chronic pain management. Implications for practice•Using a complexity informed approach (Human Learning Systems' Learning Cycles) in the management of chronic pain.•Structured and efficient way of providing a co-produced, person specific approach to practice.•Practical way of incorporating biological, psychological and social factors influencing the person's experience of chronic pain into management, without assuming causal weighting.•Epistemological implications for research and the current understanding of hierarchy of evidence.

Language-Inspired Relation Transfer for Few-Shot Class-Incremental Learning.

Depicting novel classes with language descriptions by observing few-shot samples is inherent in human-learning systems. This lifelong learning capability helps to distinguish new knowledge from old ones through the increase of open-world learning, namely Few-Shot Class-Incremental Learning (FSCIL). Existing works to solve this problem mainly rely on the careful tuning of visual encoders, which shows an evident trade-off between the base knowledge and incremental ones. Motivated by human learning systems, we propose a new Language-inspired Relation Transfer (LRT) paradigm to understand objects by joint visual clues and text depictions, composed of two major steps. We first transfer the pretrained text knowledge to the visual domains by proposing a graph relation transformation module and then fuse the visual and language embedding by a text-vision prototypical fusion module. Second, to mitigate the domain gap caused by visual finetuning, we propose context prompt learning for fast domain alignment and imagined contrastive learning to alleviate the insufficient text data during alignment. With collaborative learning of domain alignments and text-image transfer, our proposed LRT outperforms the state-of-the-art models by over 13% and 7% on the final session of miniImageNet and CIFAR-100 FSCIL benchmarks.

Theta mediated dynamics of human hippocampal-neocortical learning systems in memory formation and retrieval

Episodic memory arises as a function of dynamic interactions between the hippocampus and the neocortex, yet the mechanisms have remained elusive. Here, using human intracranial recordings during a mnemonic discrimination task, we report that 4-5 Hz (theta) power is differentially recruited during discrimination vs. overgeneralization, and its phase supports hippocampal-neocortical when memories are being formed and correctly retrieved. Interactions were largely bidirectional, with small but significant net directional biases; a hippocampus-to-neocortex bias during acquisition of new information that was subsequently correctly discriminated, and a neocortex-to-hippocampus bias during accurate discrimination of new stimuli from similar previously learned stimuli. The 4-5 Hz rhythm may facilitate the initial stages of information acquisition by neocortex during learning and the recall of stored information from cortex during retrieval. Future work should further probe these dynamics across different types of tasks and stimuli and computational models may need to be expanded accordingly to accommodate these findings.

Naturalistic Scene Modelling: Deep Learning with Insights from Biology

Advances in machine learning coupled with the abundances of training data has facilitated the deep learning era, which has demonstrated its ability and effectiveness in solving complex detection and recognition problems. In general application areas with elements of machine learning have seen exponential growth with promising new and sophisticated solutions to complex learning problems. In computer vision, the challenge related to the detection of known objects in a scene is a thing of the past. With the tremendous increase in detection accuracies, some close to that of human detection, there are several areas still lagging in computer vision and machine learning where improvements may call for more architectural designs. In this paper, we propose a physiologically inspired model for scene understanding that encodes three key components: object location, size and category. Our aim is to develop an energy efficient artificial intelligent model for naturalistic scene understanding capable of deploying on a low power neuromorphic hardware. We have reviewed recent advances in deep learning architecture that have taken inspiration from human or primate learning systems and provided direct to future advancement on deep learning with inspiration from physiological experiments. Upon a review of areas that have benefitted from deep learning, we provide recommendations for enhancing those areas that might have stalled or grinded to a halt with little or no significant improvement.

State-transition-free reinforcement learning in chimpanzees (Pan troglodytes).

The outcome of an action often occurs after a delay. One solution for learning appropriate actions from delayed outcomes is to rely on a chain of state transitions. Another solution, which does not rest on state transitions, is to use an eligibility trace (ET) that directly bridges a current outcome and multiple past actions via transient memories. Previous studies revealed that humans (Homo sapiens) learned appropriate actions in a behavioral task in which solutions based on the ET were effective but transition-based solutions were ineffective. This suggests that ET may be used in human learning systems. However, no studies have examined nonhuman animals with an equivalent behavioral task. We designed a task for nonhuman animals following a previous human study. In each trial, participants chose one of two stimuli that were randomly selected from three stimulus types: a stimulus associated with a food reward delivered immediately, a stimulus associated with a reward delivered after a few trials, and a stimulus associated with no reward. The presented stimuli did not vary according to the participants' choices. To maximize the total reward, participants had to learn the value of the stimulus associated with a delayed reward. Five chimpanzees (Pan troglodytes) performed the task using a touchscreen. Two chimpanzees were able to learn successfully, indicating that learning mechanisms that do not depend on state transitions were involved in the learning processes. The current study extends previous ET research by proposing a behavioral task and providing empirical data from chimpanzees.

Crossmodal interactions in human learning and memory.

Most studies of memory and perceptual learning in humans have employed unisensory settings to simplify the study paradigm. However, in daily life we are often surrounded by complex and cluttered scenes made up of many objects and sources of sensory stimulation. Our experiences are, therefore, highly multisensory both when passively observing the world and when acting and navigating. We argue that human learning and memory systems are evolved to operate under these multisensory and dynamic conditions. The nervous system exploits the rich array of sensory inputs in this process, is sensitive to the relationship between the sensory inputs, and continuously updates sensory representations, and encodes memory traces based on the relationship between the senses. We review some recent findings that demonstrate a range of human learning and memory phenomena in which the interactions between visual and auditory modalities play an important role, and suggest possible neural mechanisms that can underlie some surprising recent findings. We outline open questions as well as directions of future research to unravel human perceptual learning and memory.

Ethical and relational leadership in a complex world: the use of the Human Learning Systems and social pedagogical leadership framework to navigating complexity

The Covid-19 pandemic brought into sharp focus the complexities and challenges facing social work and social care organisations. Using theory, ethics and practice, this article aims to critically explore an ethical and relational approach to leadership that can help navigate complexity in social work in the UK. Explaining the emerging Human Learning System, this article argues that this paradigm can offer an alternative to the new public management, with its focus on the three core aspects of markets, matrix and management. Key to this discussion is the social pedagogy leadership framework, which aims to assist the navigation of complexity in direct practice and support positive organisational cultures and systems as discussed within the Human Learning System. The leadership framework is informed by the principles of social pedagogy, which are seen as an ethical philosophical approach to leadership and direct practice. It places human relationships and rights within the everyday and at the heart of leadership and organisational change – as Berit Bae has stated, human rights must be applied in the everyday, not just in certain situations. Using the philosophical principles and core ideas of social pedagogy and the Human Learning System, the framework is designed to support leadership, which is relational and ethical at its core, while also assisting the navigation of uncertainty and the complexities of the work undertaken by modern social work and social care organisations.

Unlocking the System: Place-based ways of working with children, their families and a neighbourhood psychologist in Bettws, Wales

In Public Health, recognition of the importance of the social determinants of health has led to significant shifts in practice and research. In the mental health sector, we are making great progress but ‘treating’ the social determinants of mental health requires a focus more on policy than on medication, therapy, and neurobiological innovation. It is now widely acknowledged that our mental health is largely determined by the conditions in which we are born, grow, work, live, and age. Conditions full of adversity and deprivation create a lack of opportunity, poor infrastructure and socially disconnected communities. These circumstances foster the conditions for community level adversity, trauma, and injustice. Solving the mental health crisis is therefore not about more access to one-to-one therapy, counselling, but about creating psychosocially health circumstances and communities for people to live in. For emotional health outcomes, we know the first 1,000 days but particularly the first two months are most crucial. It’s therefore important that support is provided early and sustained over this critical period. Working with parents with a child aged seven or younger alongside local primary schools using the human learning systems approach this project aims to explore how to communities can heal from trauma and adversity in ways that are human-rights focused, non-violent, trauma-informed, community-led, healing and culturally sensitive. Based on the adverse community experiences and resilience framework whilst using storytelling for systems change the project will look to co-develop a place-based approach to understanding and improving community wellbeing and resilience in Bettws, Newport.

Learning to Transform by Implementing AI into Administrative Decisions – Disruptive Mindset as the Key for Agility in the Crisis

The Objectives of the research paper are investigations in how a disruptive Mindset can have effects on interaction by representatives of German Authorities in times of crisis. The interdisciplinary focus of human transformative learning maybe sets linking options to machine learning approaches to speed up qualified decision making. The Approach is in addition to an introduction of implementation strategies for AI technology in the local governance of German Municipalities, what competencies of persons in charge are required to get use of technological possibilities. In this respect, the investigation bases on findings of three studies in field of “scenarios for AI in the public sector”, “AI as impact on competencies in the field of corporate finance” and “competencies in a digitalized working environment”. The Results concern the understanding what preconditions for human and AI-oriented learning systems are necessary in suitable scenarios of public sector performance. Also, convergent competence structures, based on transformative learning theories, should show a way for implementation based on models for interaction in critical situations. The Value of the paper lies in the combined View of administrative requirements in relation to learning-strategies considering technological scenarios of AI for professionalize the decision-making processes in times of Covid-19 pandemic crises.

New development: Responding to complexity in public services—the human learning systems approach

IMPACT Current approaches to public management based on principles of marketization, management and measurement are increasingly being seen to fail when faced with the complex world of public services. The Human Learning Systems (HLS) concept represents an alternative approach which embraces the complexity of the real world of organizations working to deliver services. Produced in collaboration with an emerging community of funders, managers and commissioners of services, HLS offers a framework which bridges academic complexity theory and the diverse contexts of practice. This article introduces HLS as a means to enable organizations, practitioners and service users to work together more effectively. ABSTRACT The challenges facing public services and non-profit organizations are complex and multi-faceted, confounding the orthodoxies of bureaucratic public administration and New Public Management approaches. This article discusses the merits and potential of the emerging ‘Human Learning Systems’ (HLS) approach to the funding, commissioning and management of public services as an alternative management logic. Building on prior introductory work, the authors analyse the current state of development, content and operation of HLS and its collaborative process, involving more than 300 organizations. Drawing on the experience of public and non-profit service professionals in adopting and experimenting with this approach, the authors found that HLS can provide a helpful and innovative conceptual frame to promote constructive engagement with complexity in public management theory and practice.

Integrating Machine Learning with Human Knowledge.

SummaryMachine learning has been heavily researched and widely used in many disciplines. However, achieving high accuracy requires a large amount of data that is sometimes difficult, expensive, or impractical to obtain. Integrating human knowledge into machine learning can significantly reduce data requirement, increase reliability and robustness of machine learning, and build explainable machine learning systems. This allows leveraging the vast amount of human knowledge and capability of machine learning to achieve functions and performance not available before and will facilitate the interaction between human beings and machine learning systems, making machine learning decisions understandable to humans. This paper gives an overview of the knowledge and its representations that can be integrated into machine learning and the methodology. We cover the fundamentals, current status, and recent progress of the methods, with a focus on popular and new topics. The perspectives on future directions are also discussed.

Bio-inspired Analysis of Deep Learning on Not-So-Big Data Using Data-Prototypes

Deep artificial neural networks are feed-forward architectures capable of very impressive performances in diverse domains. Indeed stacking multiple layers allows a hierarchical composition of local functions, providing efficient compact mappings. Compared to the brain, however, such architectures are closer to a single pipeline and require huge amounts of data, while concrete cases for either human or machine learning systems are often restricted to not-so-big data sets. Furthermore, interpretability of the obtained results is a key issue: since deep learning applications are increasingly present in society, it is important that the underlying processes be accessible and understandable to every one. In order to target these challenges, in this contribution we analyze how considering prototypes in a rather generalized sense (with respect to the state of the art) allows to reasonably work with small data sets while providing an interpretable view of the obtained results. Some mathematical interpretation of this proposal is discussed. Sensitivity to hyperparameters is a key issue for reproducible deep learning results, and is carefully considered in our methodology. Performances and limitations of the proposed setup are explored in details, under different hyperparameter sets, in an analogous way as biological experiments are conducted. We obtain a rather simple architecture, easy to explain, and which allows, combined with a standard method, to target both performances and interpretability.

Editorial for EAIT 2016, Issue 3

Some of you will have read the 2015 OECD report: “Students, Computers and Learning: Making the Connection” which indicates that, based on an analysis of PISA data, where ICT is used in the classroom its impact on student performance is mixed, with no appreciable improvements in student achievement in reading, mathematics or science. Is the OECD right in their assessment? Are we wasting our time with using ICT in education? How do others see this? It would be good to have some articles on this important topic in future issues of EAIT. Beginning this issue is an article from Martin Tallvid (University of Gothenburg, Sweden), titled: “Understanding teachers’ reluctance to the pedagogical use of ICT in the 1:1 classroom”. The article begins by noting the emergence of one laptop per student-initiatives (1:1) as a solution to disappointment with the pace of transformation of teaching methods in schools. Many research studies, however, have indicated that despite major expenditure, increased access and improved technical equipment, few teachers have integrated ICT in the curriculum in a way that leads to significant changes in classroom practice. The article explores teachers’ reluctance towards the pedagogical use of personal laptops in secondary schools and identifies five different, but overlapping, patterns in the explanations for their reluctance: lack of technical competence, not worth the effort, insufficient material, diminishing control and lack of time. “Politic of security, privacy and transparency in human learning systems” is an article by Adil Jeghal, Lahcen Oughdir and Hamid Tairi (University of Sidi Mohamed Ben Abdellah, Taza, Morocco). This article looks at the important role of Information Security, especially in newer education environments such as e-learning environments. The authors point out that the preservation of confidentiality has become a major issue for the majority of applications that process personal information and that the sensitivity of this information requires creators to set rules for the sharing and use of access control policies. Educ Inf Technol (2016) 21:499–502 DOI 10.1007/s10639-016-9479-8

A component-based knowledge domain model for adaptive human learning systems

Adaptive human learning systems (AHLSs) are important tools to personalise learning. However, the used domain representation formalisms lack the needed precision and flexibility those make the domains efficiently adaptable and intensively reusable. To address this issue, we propose a component based knowledge domain for an AHLS that aims to improve the adapting efficiency and provides intensive reuse of the pre-built (sub) knowledge domains. To show the feasibility and the benefits of the proposed AHLS, a prototype that experiments the explanation variants method is implemented. So, unlike the other, our AHLS achieves the adapted learning by (re)selecting and sequencing the appropriate linear combination of the component variants explaining the corresponding concepts. Also, as a more challenging task, to get a compromised solution of the conflicting learning goals and to reduce the substantial overhead, the adapting is formulated as a multi-objective component variants selection problem and it is implemented using Genetic Algorithms.

Politic of security, privacy and transparency in human learning systems

The preservation of confidentiality has become a major issue for the majority of applications that process personal information, the sensitivity of this information requires creators to set rules for the sharing and use of access control policies. A great deal of research has already been conducted in educational environments. However, one aspect that has not received much attention is the important role of Information Security, especially in newer education environments such as the e-learning environment. In this article we seek to propose a policy of security in such a way to affect views of profiles for each user to control the information accessed by this user.

Fuzzy Dissimilarity based Multidimensional Scaling and its Application to Collaborative Learning Data

Recently, the elucidation of learning activity for human learning systems has gained tremendous interests in many areas including neuroscience, brain sciences, behavioral sciences, and education. The main problems of these data are noise and the large amount data (big data). Multidimensional scaling (MDS) is well known method to capture the similarity of objects in lower dimensional configuration space and latent cognitive factors as the dimensions. How- ever, ordinary MDS is based on the Euclidean distance which often fails to capture the similarity relationship in the lower dimensional space. The main reason for this fault is that data usually does not have significant variance to be captured by the MDS. Therefore, in this study, we exploit the latent classification structure of variables to the distance and propose a new dissimilarity and a new multidimensional scaling based on this dissimilarity. We show a better performance of the proposed method by using a time series log data of mobile learning with the collaboration of several students.

Critical Roles for Anterior Insula and Dorsal Striatum in Punishment-Based Avoidance Learning

The division of human learning systems into reward and punishment opponent modules is still a debated issue. While the implication of ventral prefrontostriatal circuits in reward-based learning is well established, the neural underpinnings of punishment-based learning remain unclear. To elucidate the causal implication of brain regions that were related to punishment learning in a previous functional neuroimaging study, we tested the effects of brain damage on behavioral performance, using the same task contrasting monetary gains and losses. Cortical and subcortical candidate regions, the anterior insula and dorsal striatum, were assessed in patients presenting brain tumor and Huntington disease, respectively. Both groups exhibited selective impairment of punishment-based learning. Computational modeling suggested complementary roles for these structures: the anterior insula might be involved in learning the negative value of loss-predicting cues, whereas the dorsal striatum might be involved in choosing between those cues so as to avoid the worst.