Archives: AI in Computer division

Rotary unmanned aerial vehicles (UAVs), also known as drones, have various advantages, yet their actual applications are limited owing to their flight range. However, increasing the flight range by enhancing the hardware is a challenging task. In this study, we introduce the first step of systematic drone low-power optimization based on the framework of electronic design automation (EDA). We attempt drone power management without in-depth knowledge of aerodynamics and control theory. Instead, we introduce a novel power model of drones using physical parameters that can affect power consumption, such as the three-axis velocity and acceleration, drone height, wind velocity, and the weight and volume of payloads. We detail the experimental setup, power modeling, accuracy verification, and optimization for minimum energy paths. We achieved over 90% accuracy in power modeling without depending on aerodynamics. The proposed approach shows the feasibility of energy-aware rotary UAV flight trajectory optimization considering the external forces affecting drones such as wind. The proposed method presents up to 24.01% energy saving through path changes considering external forces.

Reference : D. Hong, S. Lee, Y. H. Cho, D. Baek, J. Kim and N. Chang, “Least-Energy Path Planning With Building Accurate Power Consumption Model of Rotary Unmanned Aerial Vehicle,” in IEEE Transactions on Vehicular Technology, vol. 69, no. 12, pp. 14803-14817, Dec. 2020, doi: 10.1109/TVT.2020.3040537.

Ki Hyun Tae, a Ph.D. student of Prof. Steven Euijong Whang in the EE department, proposed a selective data acquisition framework for accurate and fair machine learning models.

As machine learning becomes widespread in our everyday lives, making AI more responsible is becoming critical. Beyond high accuracy of AI, the key objectives of responsible AI include fairness, robustness, explainability, and more. In particular, companies including Google, Microsoft, and IBM are emphasizing responsible AI.

Among the objectives, this work focuses on model fairness. Based on the key insight that the root cause of unfairness is in biased training data, Ki Hyun proposed Slice Tuner, a selective data acquisition framework that optimizes both model accuracy and fairness. Slice Tuner efficiently and reliably manages learning curves, which are used to estimate model accuracy given more data, and utilizes them to provide the best data acquisition strategy for training an accurate and fair model.

The research team believes that Slice Tuner is an important first step towards realizing responsible AI starting from data collection. This work was presented at ACM SIGMOD (International Conference on Management of Data) 2021, a top Database conference.

For more details, please refer to the links below.

 

[논문 정보]

논문명: Slice Tuner: A Selective Data Acquisition Framework for Accurate and Fair Machine Learning Models

저자: 태기현, 황의종(지도교수)

논문 링크: https://arxiv.org/abs/2003.04549

논문 슬라이드: https://docs.google.com/presentation/d/1thnn2rEvTtcCbJc8s3TnHQ2IEDBsZOe66-o-u4Wb3y8/edit?usp=sharing

학회 발표 영상: https://youtu.be/QYEhURcd4u4?list=PL3xUNnH4TdbsfndCMn02BqAAgGB0z7cwq

Professor. Steven Euijong Whang and Changho Suh’s research team in the School of Electrical Engineering has developed a new batch selection technique for fair artificial intelligence (AI) systems. The research was led by Ph.D. student Yuji Roh (advisor: Steven Euijong Whang) and was conducted in collaboration with Professor Kangwook Lee from the Department of Electrical and Computer Engineering at the University of Wisconsin-Madison.

 

AI technologies are now widespread and influence everyday lives of humans. Unfortunately, researchers have recently observed that machine learning models may discriminate against specific demographics or individuals. As a result, there is a growing social consensus that AI systems need to be fair.

 

The research team proposes FairBatch, a new batch selection technique for building fair machine learning models. Existing fair training algorithms require significant non-trivial modifications either in the training data or model architecture. In contrast, FairBatch effectively achieves high accuracy and fairness with only a single-line change of code in the batch selection, which enables FairBatch to be easily deployed in various applications. FairBatch’s key approach is solving a bi-level optimization for jointly achieving accuracy and fairness.

 

This research was presented at the International Conference for Learning Representations (ICLR) 2021, a top machine learning conference. More details are in the links below.

Figure 1. A scenario that shows how FairBatch adaptively adjusts batch ratios in model training for fairness.

 

Figure 2.  PyTorch code for model training where FairBatch is used for batch selection. Only a single-line code change is required to replace an existing sampler with FairBatch, marked in blue.

 

 

[Paper information and links]

Title: FairBatch: Batch Selection for Model Fairness

Authors: Yuji Roh (KAIST EE), Kangwook Lee (Wisconsin-Madison Electrical & Computer Engineering), Steven Euijong Whang (KAIST EE), and Changho Suh (KAIST EE)

 

Paper: https://openreview.net/forum?id=YNnpaAKeCfx

Source code: https://github.com/yuji-roh/fairbatch

Slides: https://docs.google.com/presentation/d/1IfaYovisZUYxyofhdrgTYzHGXIwixK9EyoAsoE1YX-w/edit?usp=sharing

 

 

Heterogeneous memory-based hardware and system software framework for accelerating graph neural networks

 

Graph Neural Networks(GNN), which can infer embeddings of untrained graphs, are emerging graph machine learning algorithms that are widely used in recommendation systems, lidars, social computing, and various natural science research that process large amounts of data. Since GNNs use large-scale non-Euclidean data to learn and deduce, they have very high accuracy, versatility, and compatibility with other graph processing systems.

 

Despite the advantages, the lack of specialized acceleration systems in GNNs and the lack of hardware design have significantly hindered the accessibility and the absence of GNN accelerations. To address the limitations, research is widely performed in areas including devices, circuits, and systems. However, performing research in individual fields fails to innovatively extend and improve the range of the performance and applications of GNNs.

 

The final goal of the work is to explore and study devices, circuits, and computer architectures vertically, design non-memory/memory semiconductors that specialize in AI, and develop low-power/high-performance AI hardware-software platforms to accelerate the emerging GNNs. Three GNN core technologies include 1) accelerator framework: designing data floor-based, heterogeneous accelerated hardware accelerator control framework that includes the development of versatile programming model for graph machine learning and GNN layer placement. 2) GNN hardware acceleration: redesigning accelerated processing circuits for accelerating GNN computation and designing ReRAM-based heterogeneous core for large ReRAM-based storage classes and improvement of area/power efficiency. 3)New heterogeneous non-volatile memory devices: developing three-terminal devices ReRAM device/memory array with byte input/output, multi-level, and high-capacity non-volatile characteristics and developing reliable two-terminal ReRAM device/memory array for sequential computation.

 

The development of AI-specific non-memory/memory semiconductors and low-power/high-performance AI hardware-software platform technologies in this research is expected to dominate high-value markets through technologies of integrated AI platforms.

Our department’s professor Hyunchul Shim’s lab has won the MSIT AI Grand Challenge Competition Control Intelligence Part for the 2^nd year in a row.

This year’s competition was based on how AI can solve various problems that will emerge from the AI based future. The challenge was to set an idea about how to use AI for responding to people’s life in various areas and social issues.

Composed of 4 major areas, professor Shim’s team applied to track 4, where a total of 5 teams applied. Professor Shim’s team used the idea of how to safely use drones to deliver supplies and goods to people.

Winning for the 2^nd year in a row, Professor Shim’s team used only 100% self developed technologies (accurate positioning, high speed flight control) for the competition. Winning the competition again this year, the group will be funded a total of 2.4 billion KW which is similar to the DARPA challenge award prize. Along with the development of AI, professor Shim said that his team will further develop inside drone flight, unmanned civilian aircraft, autonomous vehicles, delivery robots and other vehicle related AI technologies.

Once again we congratulate professor Shim’s team for the reward.

Our department’s Ph.D Candidate Seung Ho Na from professor Seungwon Shin’s lab has achieved the grand award form the 4th Financial Security paper competition.

The 4th Financial Security paper competition and financial big-data idea competition awards ceremony was held last September 2nd at the Yeouido financial security education center.

A total of 38 papers were accepted and 7 of them were selected for their excellence.

The grand award (Head of Financial Council Award) was given to Ph.D Candidate Seung Ho Na’s team with the paper titled as “Research on United Learning Techniques for Guaranteeing Privacy” which proposes an AI based security scheme.

Ph.D candidate Hyeonggwon Hong’s team from the AI Graduate school professor Junmo Kim’s lab also participated in the paper competition, and due to COVID-19 Ph.D Candidate Seung Ho Na attended the awarding ceremony as the student representative.

A total of 7 million KW was rewarded to Ph.D Candidate Seung Ho Na’s team.

We once again congratulate Ph.D Candidate Seung Ho Na for the achievement on founding a new perspective on financial security.