Semantic Memory Modeling and Memory Interaction in Learning Agents

Wang, Wenwen; Tan, Ah-Hwee; Teow, Loo-Nin

doi:10.1109/tsmc.2016.2531683

Cited by 17 publications

(15 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the reinforcement learning domain, Horzyk et al [9] proposed an episodic memory architecture composed of a tree-structured memory. In a similar line of work, the authors in [10], [32], [33], [34] investigate possible interaction structures for semantic memory. However all these frameworks are proposed in the reinforcement learning domain and a substantial amount of re-engineering is required to adapt those strategies to the supervised learning domain.…”

Section: Neural Memory Networkmentioning

confidence: 99%

“…We are motivated by the neuroscience observations provided in [1]. They present strong evidence towards activations of brain areas known to be involved with perception: Episodic Memory (EM) where personal experiences are used to determine the similarities between current sensory observation and the stored experiences [8], [37]; and Semantic Memory (SM) where foundations of knowledge and concepts are stored [9], [10]. Figure 1 shows the overall structure of the proposed approach.…”

Section: Architecturementioning

confidence: 99%

“…Therefore we maintain a separate EM for each player. However it has been shown that SM incorporates extracted knowledge of a cohort of people [10]. When drawing parallels to a tennis scenario SM would represent the rules of the game, the game dynamics, etc.…”

Section: Player Specific Adaptationmentioning

confidence: 99%

“…For instance, it has been shown that expert tennis players are better at detecting events in advance [1], [7] and posses better knowledge/ expertise of situational probabilities [3]. Further investigation of human neurological structures have revealed that those capabilities occur due to a bottom-up computational process [1] within the human brain, from sensory memory to the experiences stored in episodic memory [8], [9] and knowledge derived in semantic memory [9], [10].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Memory Augmented Deep Generative models for Forecasting the Next Shot Location in Tennis

Fernando

Denman

Sridharan

et al. 2020

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

This paper presents a novel framework for predicting shot location and type in tennis. Inspired by recent neuroscience discoveries we incorporate neural memory modules to model the episodic and semantic memory components of a tennis player. We propose a Semi Supervised Generative Adversarial Network architecture that couples these memory models with the automatic feature learning power of deep neural networks, and demonstrate methodologies for learning player level behavioural patterns with the proposed framework. We evaluate the effectiveness of the proposed model on tennis tracking data from the 2012 Australian Tennis open and exhibit applications of the proposed method in discovering how players adapt their style depending on the match context.

show abstract

Section: Neural Memory Networkmentioning

confidence: 99%

Section: Architecturementioning

confidence: 99%

Section: Player Specific Adaptationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Memory Augmented Deep Generative models for Forecasting the Next Shot Location in Tennis

Fernando

Denman

Sridharan

et al. 2020

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

show abstract

“…Due to its flexibility and ease of use, TL has been well-known as a popular problem solver that enjoyed its significant success across a wide realm of real-world applications including computer vision [2], natural language processing [3], etc. Recent study on TL has also started to investigate multi-agent systems (MASs) wherein agents tend to benefit from the knowledge transferred from their partners of high payoffs, hence improving agents' performance in more efficient problem-solving [4] [5] [6]. Emerging multi-agent TL approaches include Advice Exchange (AE) mechanisms [7], a Parallel Transfer Learning Yaqing Hou is with the Data Science and Artificial Intelligence Research Centre (DSAIR), School of Computer Science and Engineering, Nanyang Technological University (NTU), Singapore 639798 (e-mail: yaqinghou@ntu.edu.sg).…”

Section: Introductionmentioning

confidence: 99%

Evolutionary Multiagent Transfer Learning With Model-Based Opponent Behavior Prediction

Hou

Ong

Tang

et al. 2021

IEEE Trans. Syst. Man Cybern, Syst.

View full text Add to dashboard Cite

This paper embarks a study on multi-agent transfer learning for addressing the specific challenges that arise in complex multi-agent systems where agents have different or even competing objectives. Specifically, beyond the essential backbone of a state-of-the-art evolutionary Transfer Learning framework (eTL), this paper presents the novel Transfer Learning framework with Prediction (eTL-P) as an upgrade over existing eTL to endow agents with abilities to interact with their opponents effectively by building candidate models and accordingly predicting their behavioral strategies. To reduce the complexity of candidate models, eTL-P constructs a monotone submodular function, which facilitates to select Top-K models from all available candidate models based on their representativeness in terms of behavioral coverage as well as reward diversity. eTL-P also integrates social selection mechanisms for agents to identify their better performing partners, thus improving their learning performance and reducing the complexity of behavior prediction by reusing useful knowledge with respect to their partners' mind universes. Experiments based on a partner-opponent minefield navigation task (PO-MNT) have shown that eTL-P exhibits the superiority in achieving higher learning capability and efficiency of multiple agents when compared to the state-of-the-art multiagent transfer learning approaches.

show abstract

ART neural network-based integration of episodic memory and semantic memory for task planning for robots

Nasir

Kim

2019

Auton Robot

View full text Add to dashboard Cite

Automated task planning for robots faces great challenges in that the sequences of events needed for a particular task are mostly required to be hard-coded. This can be a cumbersome process, especially, when the user wants a robot to learn a large number of similar tasks with different objects that are semantically related. We propose a novel approach of user preferencebased integrated multi-memory model (pMM-ART). This approach focuses on exploiting a semantic hierarchy of objects alongside an episodic memory for enhancing the behavior of an autonomous agent. We analyze the functioning principle of the proposed model by teaching it a few distinct domestic tasks and observe that it is able to carry out a large number of similar tasks based on the semantic similarities between learned objects. We also demonstrate, via experiments using Mybot, our ability to reach those goals that are not possible without the integration of semantic knowledge with episodic memory.

show abstract

Semantic Memory Modeling and Memory Interaction in Learning Agents

Cited by 17 publications

References 27 publications

Memory Augmented Deep Generative models for Forecasting the Next Shot Location in Tennis

Memory Augmented Deep Generative models for Forecasting the Next Shot Location in Tennis

Evolutionary Multiagent Transfer Learning With Model-Based Opponent Behavior Prediction

ART neural network-based integration of episodic memory and semantic memory for task planning for robots

Contact Info

Product

Resources

About