Learning Process Of Tasks Research Articles

A generalization of multi-source fusion-based framework to stock selection

Selecting outstanding technology stocks for investment is challenging. Specifically, the research of the investment for academic purposes is not mature enough due to the disarray of the publications and the overwhelming informative experience from profit-making websites. Often, some authors entitle the stock price prediction as a stock selection problem; both prediction and selection are just sub-sections of portfolio management. Moreover, stock websites provide numerous potential criteria showing various evaluations to simulate and monitor the stock market professionally, which increases the difficulty of academic studies on stock selection.The paper generalizes a novel framework with a user-interactive interface for stock selection problems based on multi-source data fusion and decision-level fusion to enhance reliability limited by the narrow criteria performance and the strength of a model overcoming the weakness of a single-performed model. This framework benefits the time-series prediction and decision-making study. Besides, we propose adopting dynamic time warping to assist a task-learning process by customizing a loss function that improves the accuracy of data prediction. The experiment shows that the proposed method reduces the prediction log error by 6.3% on average and decreases the warping cost by 5.6% on average over all cases of real-situation data. Finally, we illustrate the proposed framework by implementing it in a real-world stock data selection. The results are practical and effective, further justified through a detailed ablation study. The source code will be available at https://github.com/lingping-fuzzy.

Auditory Cortex Neurons Show Task-Related and Learning-Dependent Selectivity toward Sensory Input and Reward during the Learning Process of an Associative Memory Task.

The activity of primary auditory cortex (A1) neurons is modulated not only by sensory inputs but also by other task-related variables in associative learning. However, it is unclear how A1 neural activity changes dynamically in response to these variables during the learning process of associative memory tasks. Therefore, we developed an associative memory task using auditory stimuli in rats. In this task, rats were required to associate tone frequencies (high and low) with a choice of ports (right or left) to obtain a reward. The activity of A1 neurons in the rats during the learning process of the task was recorded. A1 neurons increased their firing rates either when the rats were presented with a high or low tone (frequency-selective cells) before they chose either the left or right port (choice-direction cells), or when they received a reward after choosing either the left or right port (reward-direction cells). Furthermore, the proportion of frequency-selective cells and reward-direction cells increased with task acquisition and reached the maximum level in the last stage of learning. These results suggest that A1 neurons have task- and learning-dependent selectivity toward sensory input and reward when auditory tones and behavioral responses are gradually associated during task training. This selective activity of A1 neurons may facilitate the formation of associations, leading to the consolidation of associative memory.

A hybrid ambipolar synaptic transistor emulating multiplexed neurotransmission for motivation control and experience-dependent learning

Artificial synapses with full synapse-like functionalities are of crucial importance for the implementation of neuromorphic computing and bioinspired intelligent systems. In particular, the development of artificial synapses with the capability to emulate multiplexed neural transmission is highly desirable, but remains challenging. In this work, we proposed a hybrid ambipolar synaptic transistor that combines two-dimensional (2D) molybdenum disulfide (MoS2) sheet and crystalline one-dimensional (1D) poly(3-hexylthiophene-2,5-diyl) polymer nanowires (P3HT NWs) as dual excitatory channels. Essential synaptic functions, including excitatory postsynaptic current, paired-pulse facilitation, synaptic potentiation and depression, and dynamic filtering were emulated using the synaptic transistor. Benefitting from the dual excitatory channels of the synaptic transistor, the device achieved a fast switch between short-term and long-term memory by altering the charge carriers in the dual channels, i.e., electrons and holes. This emulated the multiplexed neural transmission of different excitatory neurotransmitters, e.g., dopamine and noradrenaline. The plasticity-switchable artificial synapse (PSAS) simulates the task-learning process of individuals under different motivations and the impact of success or failure on task learning and memory, which promises the potential to enable complex functionalities in future neuromorphic intelligent electronics.

Decentralized Multi-Agent Control of a Manipulator in Continuous Task Learning

Many real-world tasks require multiple agents to work together. When talking about multiple agents in robotics, it is usually referenced to multiple manipulators in collaboration to solve a given task, where each one is controlled by a single agent. However, due to the increasing development of modular and re-configurable robots, it is also important to investigate the possibility of implementing multi-agent controllers that learn how to manage the manipulator’s degrees of freedom (DoF) in separated clusters for the execution of a given application (e.g., being able to face faults or, partially, new kinematics configurations). Within this context, this paper focuses on the decentralization of the robot control action learning and (re)execution considering a generic multi-DoF manipulator. Indeed, the proposed framework employs a multi-agent paradigm and investigates how such a framework impacts the control action learning process. Multiple variations of the multi-agent framework have been proposed and tested in this research, comparing the achieved performance w.r.t. a centralized (i.e., single-agent) control action learning framework, previously proposed by some of the authors. As a case study, a manipulation task (i.e., grasping and lifting) of an unknown object (to the robot controller) has been considered for validation, employing a Franka EMIKA panda robot. The MuJoCo environment has been employed to implement and test the proposed multi-agent framework. The achieved results show that the proposed decentralized approach is capable of accelerating the learning process at the beginning with respect to the single-agent framework while also reducing the computational effort. In fact, when decentralizing the controller, it is shown that the number of variables involved in the action space can be efficiently separated into several groups and several agents. This simplifies the original complex problem into multiple ones, efficiently improving the task learning process.

Acquisition of a spatial navigation task in the padding pool induces an increase of GABA level in the hippocampus of Swiss mice

The balance between excitatory and inhibitory glutamatergic and GABAergic systems, respectively, is crucial for the maintenance of complex cognitive functions such as learning. Using Swiss mice as experimental model, the aims of the present study were to evaluate cognitive performance in a shallow water maze (SWM) and the effects of training in this spatial navigation task on hippocampal GABA and glutamate levels. In addition, correlations between neurochemical and behavioural data, and between glutamate and GABA levels were assessed. Forty-six three-month-old mice were divided into three groups: Learning, n=18: animals submitted to the SWM task; Active, n=14: animals exposed to the SWM, without the demand of performing a cognitive task and Control, n=14: the animals were kept in the vivarium without contact with the SWM. There was significant training effect indicating that the Learning group animals have learned the task. Regarding neurochemical data, the findings of the present work show for the first time that the task learning process in SWM has a significant effect on GABA levels in the hippocampus. The relationship between the two neurotransmitters, observed in the control animals, was adjusted by a significant increase in hippocampal GABA levels caused by the spatial training performed by the animals from the Learning group. However, the relationship observed in control condition is disrupted by a subsequent exposure to the maze in the absence of a spatial cognitive demand, as was the case of the Active group. These data open new perspectives to explore the involvement of the inhibitory and excitatory systems in the molecular mechanisms associated with different types and steps of learning processes.

Interactively shaping robot behaviour with unlabeled human instructions

In this paper, we propose a framework that enables a human teacher to shape a robot behaviour by interactively providing it with unlabeled instructions. We ground the meaning of instruction signals in the task-learning process, and use them simultaneously for guiding the latter. We implement our framework as a modular architecture, named TICS (Task-Instruction-Contingency-Shaping) that combines different information sources: a predefined reward function, human evaluative feedback and unlabeled instructions. This approach provides a novel perspective for robotic task learning that lies between Reinforcement Learning and Supervised Learning paradigms. We evaluate our framework both in simulation and with a real robot. The experimental results demonstrate the effectiveness of our framework in accelerating the task-learning process and in reducing the number of required teaching signals.

The influence of user interface design on task performance and situation awareness in a 3-player diner's dilemma game

To understand the influence of user interface on task performance and situation awareness, three levels of user interface were designed based on the three-level situation awareness model for the 3-player diner’s dilemma game. The 3-player diner's dilemma is a multiplayer version of the prisoner's dilemma, in which participants play games with two computer players and try to achieve high scores. A total of 117 participants were divided into 3 groups to participate in the experiment. Their task performance (the dining points) was recorded and their situation awareness scores were measured with the Situation Awareness Global Assessment Technique. The results showed that (1) the level-3 interface effectively improved the task performance and situation awareness scores, while the level-2 interface failed to improve them; (2) the practice effect did exist in all three conditions; and (3) the levels of user interface had no effect on the task learning process, implying that the learning rules remained consistent across different conditions.

Motor Learning Process of Virtual Reality Tasks with Different Demands in Post Stroke Individuals

Practicing tasks in Virtual Reality (VR) environment has shown to be effective for gain in cognitive and motor functions in post stroke individuals. However, there are few information about how the motor learning process of tasks with different demands occurs in this population, and if there are differences when compared with healthy individuals. The aim of the present study was to investigate the motor learning process of eight VR’s tasks with different cognitive and motor demands, in post stroke individuals (Experimental Group-EG), and compare it with healthy individuals (Control Group-CG). Eighteen individuals participated in the study performing eight sessions of practice, four for each task. The study involved: pre-test (1 session), acquisition phase (4 sessions for each task), post-test (1 session), and retention test (1 session after thirty days since the end of acquisition phase). The statistical analysis was run by a 2 x 3 ANOVA (groups x learning tests) with Tukey post hoc. At the end, EG group was able to learn three tasks, while CG learned four, one more than the ones learned by the post-stroke individuals. Task’s complexity affected, in more intensely the motor learning process in post-stroke individuals that with eight sessions were able to learn only low-complexity tasks.

THE DEVELOPMENT OF OPTIMIZATION METHODS FOR KNOWLEDGE BASE ENRICHMENT PROCESSES

The paper presents the concept of approach to the research and evaluation of the processes of intellectual activity associated with the enrichment of the knowledge base. A feature of the research of the process dynamics is the need of simultaneous consideration of such diverse factors as the complexity of information perception, the presence of the deviations of the response from the standard in the process of reproduction and accounting of the test time. A significant influence on the methods of optimization of the knowledge base enrichment process is exerted by a considerable duration of the task learning process. This causes the use of the multifactor experimental design theory to accelerate the progress towards the optimum. The research results can be used in the development of technologies for efficient knowledge assimilation, automation of skills, and also in the development of expert systems for diagnostics of the processes of intellectual activity.

The use of dynamic work sharing production methods to reduce the impact of labour turnover in serial assembly lines

Labour turnover is one of the major complications that industry faces on a daily basis. Turnover is particularly harmful to serial assembly lines. The learning process that a new operator must undergo when first introduced to an assembly line station generates blockage and starvation in the previous and successive stations, causing a reduction in the line’s throughput. In this paper, we analyse different work allocation strategies for serial assembly line designs in order to determine which designs perform best under the presence of labour turnover and a task-learning process. In particular, we compare the traditional balanced line with two other assembly line designs, a variant of the bucket brigades (BB) and a hybrid method – the modified work sharing (MWS) method. By making a judicious use of control buffers, the MWS method seeks to combine the positive characteristics of both the BB and traditional balanced lines to naturally adapt to different levels of labour turnover. The results presented in this paper show that the MWS outperforms the BB and the balanced designs.