講者：李亞棟院士，化學部

講題：Metallic Nano, Cluster, and Single Atom Catalysis

摘要：Metallic nanocrystals, clusters, and single atoms are important catalysts that have been widely used in both heterogeneous and homogeneous catalysis. It is of great significance to understand the essence and laws of catalysis for the frontier of chemistry and chemical engineering. More and more attention has been paid to the correlation of catalyst structure and catalytic properties. To date, researchers have uncovered many important factors governing the catalytic performance of nano, cluster, and single atom catalysts, including composition, shape, size, surface/interface effects and so on. However, challenges remain in the following aspects: intrinsically what the active site or center of the catalyst is in heterogeneous catalysis; which is the key factor determining the catalytic performance: composition, structure (shape), size, interaction between catalyst and support, or surface/interface effects; whether or not the latest achievements of nanotechnology could be applied to the fundamental and applied research of catalysis, so as to improve the activity, selectivity and stability of the catalyst, and if so, how to manage; whether enzyme catalysts could be manufactured with high activity and specific selectivity in ambient conditions as enzyme in nature. Herein, we made some explorative attempts and achieved valuable results in combination with several catalysis in chemical engineering.

REFERENCES
1, Yadong Li, et al. Sci. Adv., 2017, 3, e1603068.
2, Yadong Li, et al. J. Am. Chem. Soc., 2017, 139, 9795-9798.
3, Yadong Li, et al. J. Am. Chem. Soc., 2017, 139, 10976-10979.
4, Yadong Li, et al. J. Am. Chem. Soc., 2017, 139, 7294-7301.
5, Yadong Li, et al. J. Am. Chem. Soc., 2017, DOI: 10.1021/jacs.7b10194.
6, Yadong Li, et al. J. Am. Chem. Soc., 2017, 139, 9419-9422.
7, Yadong Li, et al. J. Am. Chem. Soc., 2017, 139, 8078-8081.
8, Yadong Li, et al. Angew. Chem. Int. Ed., 2017, 56, 6937-6941.
9, Yadong Li, et al. Angew. Chem. Int. Ed., 2017, 56, 11971-11975.
10, Yadong Li, et al. Angew. Chem. Int. Ed., 2017, DOI: 10.1002/anie.201710599.
11, Yadong Li, et al. Angew. Chem. Int. Ed., 2017, DOI: 10.1002/anie.201709803.
＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿

講者：陳曉非院士，地學部

講題：計算地震學及其在防震減災領域的應用
　　　Computational Seismology and Applications in Hazards Mitigation

摘要：在經濟高速發展、城鎮化進程不斷加快的今天，地震的威脅及造成的損害也越來越大。面對嚴峻的震情形勢，如何有效地減輕破壞性地震所造成的人員傷亡和經濟損失，是我國及其他地震多發國家所面臨的迫切需要解決的問題。充分了解地震斷層系統的動力學演化過程和規律、震源物理機制、地震波傳播規律以及目的地區域的場地回應等是揭示地震災害規律的關鍵。計算地震學旨在基於前沿的演算法和先進的計算設備，研究並解決這些地震科學問題。在過去十年裡，中國地震學家在這一研究領域取得了一系列重要研究成果，促進了防震減災工作的發展。
The threat and damages caused by earthquakes are becoming more and more serious as the highly developing society. How to effectively mitigate seismic disaster is an urgent problem to be solved for China as well as for other places where earthquakes frequently happen. Deeply understanding the evolution of earthquake fault system, seismic source mechanism, seismic wave propagation and so on, are crucial bases to mitigate the seismic hazard. Computational seismology aims at to uncover these basic scientific problems by using the state-of-art algorithm and computing infrastructure. In the past decades, Chinese seismologists had made significant contributions to this field, and promoted the advances in seismic hazards mitigation.
＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿

講者：徐宗本院士，信息技術科學部

講題：Model Driven Deep Learning

摘要：Deep learning (DL) has becoming a powerful, standard AI technology which helps to yield increasingly breakthroughs of learning system applications. As a representative of data driven approach, it faces however many challenges like contradictions between standardization and personalization, versatility and efficiency, the difficulties in design, anticipation and explanation for the results, and the serious dependence upon the amount and quality of training samples. On the other hand, the model-driven approach provides another learning paradigm that bases on the physical mechanism and prior modeling, which has the characteristics of determinacy and optimality while meets with obstacle of impossibility of precise modeling. In this talk we propose and formalize a data & model dual-driven learning approach, which define then the model driven deep learning (MDDL). The model driven deep learning start with construction of a Model Family (MF), which is a rough description of solution of the problem under consideration, followed then by the design of an Algorithm Family (AF) which is a collection of iterations whose limit give the solution of the model family. The Algorithm Family then unfolded into Deep Architecture (DA) with which learning can be performed. We provide examples to substantiate the effectiveness and superiority of the MDDL over others. We particularly show the following advantages of MDDL: It recedes the requirement for precise modeling in model-driven learning, provides the sound methodology for the DL network design, making it easy to incorporate into prior knowledge to make DL more efficient, designable, predictable and interpretable, and also significantly reduce the number of samples needed for DL training. Based on this study, we conclude that MDDL has great potential in the future DL research and applications.