CV

Education

• MSc in Data Science, Nazarbayev University, Kazakhstan
  Aug 2022 – Jun 2024
  GPA: 3.59 / 4.0
  Thesis: An Exploration of Video Transformers for Few-Shot Action Recognition

• BS in Computer Science, Nazarbayev University, Kazakhstan
  Aug 2018 – Jun 2022
  GPA: 3.71 / 4.0 (CS Courses GPA: 3.82 / 4.0)
  Honors: Diploma of Honor

Research Interests

Deep Learning, Computer Vision, Few-Shot Learning, Vision-Language Models.

Publications

1. Efficient Facial Expression Recognition Framework Based on Edge Computing
   Journal of Supercomputing (SJR Q2), Jan 2024
   Aikyn, N. (First Author), Zhanegizov, A., Aidarov, T., Bui, D. M., & Tu, N. A.
   DOI: 10.1007/s11227-023-05548-x

2. FedFSLAR: A Federated Learning Framework for Few-shot Action Recognition
   2024 IEEE WACV Conference, Jan 2024
   Tu, N. A., Abu, A., Aikyn, N., Makhanov, N., Lee, M. H., Le-Huy, K., & Wong, K. S.
   DOI: 10.1109/WACVW60836.2024.00035

3. Few-shot Action Recognition with Video Transformer
   2023 IEEE SITIS Conference, Nov 2023
   Aikyn, N. (First Author), Abu, A., Zhaksylyk, T., & Tu, N. A.
   DOI: 10.1109/SITIS61268.2023.00027

4. Benchmarking Federated Few-shot Learning for Video-based Action Recognition
   Under Revision
   Tu, N. A., Aikyn, N., Makhanov, N., Abu, A., Wong, K. S., & Lee, M. H.

5. Improving Vision-Language Models with Attention Mechanisms for Aerial Video Classification
   Under Revision
   Tu, N. A., Aikyn, N.

Research Experience

• Research Assistant
  Nazarbayev University, Kazakhstan
  Jul 2022 – Present

  Contributed to the project: Edge-assisted activity recognition using skeletal representation and deep learning for video surveillance.

  Focus areas: Deep Learning, Computer Vision, Few-Shot Learning, Federated Learning, Human Action Recognition, Facial Expression Recognition, Edge Computing, Vision Transformers, Vision-Language Models.

  Worked on tasks:

  1. Vision-Language Models for Aerial Video Classification
     Jun 2024 - present
     • Adapted vision-language models (e.g., CLIP) to aerial video datasets (ERA, UAV-Human, MOD20), enhancing frame sampling, prompt generation, and temporal aggregation for action detection.
     • Conducted extensive testing in zero-shot, few-shot, and fully supervised settings, with an ablation study to optimize methods, achieving SOTA results on benchmark datasets.
  2. Federated Few-Shot Action Recognition
     Aug 2023 - Jul 2024
     • Contributed to integrating few-shot learning in federated learning frameworks. Conducted experiments with meta-learning techniques (e.g., Prototypical Networks, Reptile) on MViT, Slow-Fast, and R(2+1)D models under both IID and non-IID federated settings.
     • Conducted extensive testing in zero-shot, few-shot, and fully supervised settings, with an ablation study to optimize methods, achieving SOTA results on benchmark datasets.
  3. Few-Shot Learning for Action Recognition with Video Transformers
     Jan 2023 - Nov 2023
     • Developed a few-shot action recognition framework with Transformer-based backbones (TimeSformer, Swin Transformer, MViT) and ProtoNet for metric-based meta-learning.
     • Implemented self-supervised pretraining on custom datasets and leveraged pretraining on large datasets to boost transformer model performance in 1-shot and 5-shot tasks.
  4. Transformer-Based Multi-Stream Human Action Recognition
     Jul 2022 - Jan 2023
     • Built multi-stream models integrating RGB, optical flow, and skeletal data via feature-level fusion, achieving near-SOTA results through transfer learning and keyframe sampling.
  5. Facial Expression Recognition (FER) on Edge Systems
     Jan 2022 - Nov 2022
     • Designed a FER framework optimized for edge devices, combining deep learning and edge computing for low-latency, privacy-preserving emotion recognition.
     • Achieved real-time FER with landmark-based feature engineering and lightweight models, matching SOTA accuracy on resource-constrained devices.

Achievements and Activities

• Presented paper Few-shot Action Recognition with Video Transformer at SITIS 2023 in Bangkok, Thailand, Nov 2023
• Third Prize at the International Mathematics Competition, Blagoevgrad, Bulgaria, Aug 2019.
• Gold Medal in the SIAM Mathematics Olympics, Nazarbayev University, Feb 2019.

Teaching Assistant Experience

• CSCI 390 - Artificial Intelligence, Nazarbayev University
  Fall 2022, 2023, 2024
  Assisted in proctoring exams and grading.

• CSCI 152 - Data Structures and Performance, Nazarbayev University
  Spring 2024
  Managed lab sessions and graded assignments and exams.

Skills

• Programming Languages: Python, Java, C++, C, C#, Lua, Prolog, Javascript
• Frameworks & Tools: PyTorch, OpenCV, TensorFlow, Keras, SciKit-Learn, MySQL, Linux, Git
• Languages: English (Proficient, IELTS band 7.5), Kazakh (Native), Chinese (Native)