A5052185411- AI in Service Interactions
- Speech and dialogue systems
- Social Robot Interaction and HRI
- Reinforcement Learning in Robotics
- Evolutionary Algorithms and Applications
- Software Engineering Research
- Teaching and Learning Programming
- Topic Modeling
- Metaheuristic Optimization Algorithms Research
Nazarbayev University
2021-2023
The application of Reinforcement Learning (RL) as an emergent field Machine has shown positive results in interdisciplinary fields. Although research proven its effectiveness language education through various agents (e.g., chatbots, robots, talking avatars), letter acquisition is relatively new. In light the alphabet transition from Cyrillic to Latin for Kazakh language, potential challenges might be associated with learning and memorizing new alphabet. Specifically, students extant...
Reinforcement learning has gained significant interest in modern industries for its advancements tackling challenging control tasks compared to rule-based programs. However, the robustness aspect of this technique is still under development, limiting widespread adoption. This problem become more pronounced as users switch training simulations reduce costs, resulting a reality gap that negatively affects real-world performance. One popular method employed mitigate randomizing uncertain...
This works addresses the lately initiated Cyrillic-to-Latin alphabet shift in Kazakhstan that may bring challenges for early literacy development and acquisition; both public scientific communities agree on possible resistance to acquiring using a new alphabet. To support acquisition of Kazakh Latin its handwriting, this study proposes reinforcement-learning (RL) system named QWriter. It comprises humanoid robot NAO, tablet with stylus, an RL agent learns from child's mistakes progresses...
One of the serious problems in Reinforcement Learning (RL) algorithms is that their performance usually varies when same experiment repeated or reproduced. Although RL results are hard to reproduce due algorithms' intrinsic variance, which was not investigated systematically. Through this case study on Flappy Bird environment, we introduce and characterize four important factors inconsistency algorithms: 1) level environment randomness, 2) order action-value updates process, 3) exploration...
In Kazakhstan, the ongoing Cyrillic-to-Latin alphabet shift raises challenges for early literacy development and acquisition in Kazakh language. This paper proposes QWriter system to help young children learn Latin-based its handwriting. The consists of a humanoid robot NAO, tablet with stylus, Reinforcement Learning (RL) agent that learns child's mistakes progress maximize learning shortest period time by adapting order practice words according mistakes. To evaluate effectiveness system, we...
Robot-assisted language learning produces comparable results to human tutors in a long-term study with elementary school children.
The present study applies a novel Reinforcement Learning-based (RL) alphabet learning system named QWriter for the acquisition of Kazakh Latin alphabet. We conducted between-subject design experiment with 108 children aged 6-8 years old in public school and compared their rates across two conditions: an RL-based robot human tutor (HT) as baseline. results show that learned significantly more letters HT to robot, showing is not effective short term. Yet, we observe some interesting by...