先進應用人工智能實驗室 a3ilab

Advanced Applied Artificial Intelligence Laboratory (AAAI Lab)

Welcome to the Advanced Applied Artificial Intelligence (AAAI) Laboratory. In this era where artificial intelligence (AI) stands at the forefront of technological advancements, our laboratory seeks to delve deeper into its intricacies, potentials, and challenges.

Our Core Objectives:

Development of Intelligent Systems: At the AAAI Laboratory, our primary focus lies in crafting intelligent systems that are not just theoretically advanced but practically applicable in diverse domains. From healthcare to logistics, our projects aim to revolutionize sectors through AI-driven innovations.

1. Safety First: As pioneers in the field, we recognize the imperative nature of developing AI systems that are inherently safe. This means systems that are free from biases, able to function reliably in various conditions, and resistant to adversarial attacks. Our research methodologies and protocols emphasize the safety of AI applications to ensure they contribute positively without unintentionally causing harm or discrepancies.

2. Creating Helpful Solutions: Our goal is not merely to innovate, but to do so in ways that cater directly to real-world needs. We're driven by a vision where AI serves as a helping hand to humanity, simplifying complex tasks, offering insights that were previously inconceivable, and making the world a more connected and informed space.

We constantly challenge our paradigms, remain updated with global trends, and push the boundaries of what AI can achieve. Together, let's shape the future of artificial intelligence, ensuring it's both safe and incredibly helpful for generations to come.

Research Topics:

We specialize in exploring and developing cutting-edge artificial intelligence technologies, focusing on three key research areas:

1. Niche Generative AI: Our work in niche generative AI involves developing techniques that create high-quality content in specific domains or applications. We investigate how generative models can be harnessed to produce valuable data and insights, driving innovation across various fields, including text, image, and multimedia content generation.

2. Trust-Based Discriminative AI: In the realm of trust-based discriminative AI, we focus on creating systems that can quantify uncertainty in decision-making processes. Our goal is to develop intelligent systems capable of opting to abstain from making decisions in uncertain situations, ensuring that all decisions made are reliable. This approach aims to enhance system performance and trustworthiness in complex environments.

3. Energy-Efficient Multimodal AI: Our research in energy-efficient multimodal AI addresses how AI systems can be more energy-efficient. We explore techniques such as model miniaturization, quantization, and multimodal collaboration to tackle the energy demands associated with widespread AI deployment. Our objective is to enable the sustainable operation of AI systems, supporting the future development of intelligent technologies.

Recent publications

1. Ma, C. & Liu, Y.. (2023). DeepGD3: Unknown-Aware Deep Generative/Discriminative Hybrid Defect Detector for PCB Soldering Inspection. Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 216:1326-1335 Available from https://proceedings.mlr.press/v216/ma23a.html.
2. Lee, S. P., Kini, N. P., Peng, W. H., Ma, C. W., & Hwang, J. N. (2023). Hupr: A benchmark for human pose estimation using millimeter wave radar. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 5715-5724).
3. Yang, Ting‐Ting, Ching‐Wen Ma, Jyun‐Wei Jhou, Yu‐Ting Chen, and Cheng‐Che E. Lan. "Response predictor for pigment reduction after one session of photo based therapy using convolutional neural network: A proof of concept study." Photodermatology, Photoimmunology & Photomedicine (2023).
4. J. -Y. Liao, J. -W. Hsieh and C. -W. Ma, "Automatic Meter Reading Based on Bi-Fusion MSP Network and Carry-Out Rechecking," in IEEE Access, vol. 10, pp. 96710-96719, 2022, doi: 10.1109/ACCESS.2022.3201235.
5. S. -M. Kao, C. -W. Ma and J. -W. Hsieh, "Integrating Low-Cost LiDAR and Stereo Camera for 3D Object Detection with Lidar-Stereo Fusion and Cost Volume Enhancement," TENCON 2022 - 2022 IEEE Region 10 Conference (TENCON), Hong Kong, Hong Kong, 2022, pp. 1-6, doi: 10.1109/TENCON55691.2022.9977732.
6. Y. Liu and C. -W. Ma, "Improving Tiny YOLO with Fewer Model Parameters," 2021 IEEE Seventh International Conference on Multimedia Big Data (BigMM), Taichung, Taiwan, 2021, pp. 61-64, doi: 10.1109/BigMM52142.2021.00017.

實驗室公約

1. 私人桌面，宜整理整潔。
2. 私人物品，宜收入私人使用之櫃子中。
3. 生活垃圾，宜於離開實驗室時，帶走並丟棄。

實驗室守則

1. 私人物品須放置於私人區域。
2. 離開實驗室時，椅子與移動櫃須回復定位。
3. 設置周值日生，負責保持公共區域整潔，須定位公共設備與桌椅。
4. 禁止於實驗室玩電玩，喝酒，打球，大聲喧嘩。
5. 禁止放置與研究無關之大中型私人設備，如，遊戲機，吉他等。
6. 禁止私人物品，長期放置於公共櫃子裡，如，超過一周。
7. 設置周值日生，負責公共櫃子管理，須檢查是否有私人用品占用。
8. 實驗室有非實驗用途需求時，須向老師提出申請。