The AI project on smart manufacturing would like to investigate practical problems and challenges derived from our recent industrial collaboration with high-tech manufacturing companies. Although there were some manufacturing process problems that can be overcome by conventional machine learning approaches, these problems, however, had easy-to-retrieve features. As the high-tech manufacturing process ” getting increasingly complicated, the “key” processes have become a serious challenge for most of the high-tech manufacturing.”We first take into consideration the lithography process in the semiconductor industry as the short-term goal to elaborate the artificial intelligence optimization applications. Apart from most advanced process control systems that used statistical measured data, we further attempt to make use of real data-driven approaches.(more details)
Modern technologies such as Global Positioning Systems (GPS) and mobile communication have contributed to the development of dynamic navigation planning based on real-time information. However, traffic conditions vary enormously and unpredictable accidents significantly affect planned routes, which increase the problem complexity, even though current navigation systems use information about road distances and speed limits to find the fastest routes. Therefore, online decision-making strategies play an important role in solving traffic congestion problems. We consider an old online route planning problem, called the Canadian Traveller Problem (CTP), which finds practical applications in designing dynamic navigation systems. We study several generalizations of the CTP and propose deterministic algorithms with theoretical competitive ratio. We also present the first polynomial time randomized algorithm that surpasses the deterministic lower bound. Recently, we consider the electric vehicle (EV) routing problem that takes into consideration of possible battery charging or swapping operations. We develop efficient algorithms which can be implemented on an online EV routing map interface (more details).
A fundamental goal of biology is to understand the cell as a system of interacting components and especially, almost every biological process is mediated by a network of molecular interactions. In particular, there has been a considerable amount of research devoted to the discovery and exploration of interactions between proteins in the last decade. Since many cellular activities are a result of protein interactions, proteins often interact with other proteins to perform their functions, and form a complex biological system, i.e., a protein-protein interaction (PPI) network. This powerful way of representing and analyzing the vast corpus of PPI data describes the interaction relationship among proteins in a cell. Furthermore, knowledge about the topology of a PPI network in one organism can yield insights about not only the networks of similar organisms, but also the function of their components. Hence comparison between protein interaction networks is becoming central to systems biology. We have collaborated with the MIT team and developed global alignment algorithms for performing comparative analysis of multiple biological networks (more details).
When the global energy crisis and related issues become critically important, more researchers focus on the energy management problems and especially, Smart Grid is one of the most popular research topics. In order to solve the technical challenges of communications between power plants and stations, power companies have to observe the real-time state of a power grid and continually monitor the whole electricity system. The PMU (phasor measurement unit) was invented and such devices can measure the electrical waves on a power grid and determine the health of the utility system. We consider the power observation problem of optimally placing PMU devices on wide-area power grids according to different objectives, while maintaining the ability to observe electricity systems (more details).
With the rapid growth of international logistics market, one of the most important research issues is designing a large-scale distribution network. The question of large-scale distribution network design is also becoming central to globalization supply chain management. In general, the location and network design problems have become more important and have been studied extensively during the last decade. In order to deal with different real-world applications in which the constraints and requirements appear in different scenarios, these problems can be formulated in various ways. We study capacitated facility location in large-scale networks and its application to distribution network design. In a distribution network, each distribution center or client has associated with a demand, and each plant or facility has a capacity that specifies the maximum service the plant can provide to its distribution centers. (more details).
The bin packing problem is one of the classic NP-hard combinatorial optimization problems. Given a set of n items with positive sizes, the objective is to find a packing in bins of equal capacity to minimize the number of bins required. The problem finds obvious practical usage in many industrial applications, such as the container loading problem and the cutting stock problem. There are also many variations of the bin packing problem, such as the strip packing, square packing and rectangular box packing problems. We consider the three-dimensional orthogonal bin packing problem and present new lower bounds for the problem from a combinatorial point of view. In particular, we demonstrate that the bounds theoretically dominate all previous results from the literature. The comparison is also done concerning asymptotic worst-case performance ratios. The new lower bounds can be more efficiently computed in polynomial time. In addition, we study the non-oriented model, which allows items to be rotated (more details).
Selected Publication List:
1. Erik Demaine, Yamming Huang, Chung-Shou Liao and Kunihiko Sadakane. Canadians Should Travel Randomly, in Proc. the 41st International Colloquium on Automata, Languages, and Programming (ICALP 2014), Copenhagen, Denmark. (The full version entitled Approximating the Canadian Traveller Problem with Online Randomization, is published in Algorithmica, (2020). https://doi.org/10.1007/s00453-020-00792-6)
2. Cheng-Yu Ma, Yi-Ping Phoebe Chen, Bonnie Berger , Chung-Shou Liao. Identification of Protein Complexes by Integrating Multiple Alignment of Protein Interaction Networks, Bioinformatics, Vol. 33(11), (2017), pp. 1681-1688. (DOI: 10.1093/bioinformatics/btx043)
3. Chung-Shou Liao, Shang-Hung Lu, and Zuo-Jun Max Shen. The Electric Vehicle Touring Problem, Transportation Research Part B: Methodological, Vol. 86, (2016), pp. 163-180.
4. Chung-Shou Liao and D. T. Lee. Power domination in circular-arc graphs, Algorithmica, Vol. 65 No. 2 (2013) pp. 443-466.
5. Chung-Shou Liao, Kanghao Lu, Michael Baym, Rohit Singh, and Bonnie Berger. IsoRankN: Spectral methods for global alignment of multiple protein networks, in Proc. the 17th International Conference on Intelligent Systems for Molecular Biology (ISMB 2009), Stockholm, Sweden (also invited to be published in Bioinformatics, Vol 25 No. 12 (2009) pp. i253-i258).