Invitation to two lectures by leading experts from Purdue University

Prof. Alejandra J. Magana's presentation will address teaching and learning through computation, modeling and simulation. He will summarize specific difficulties encountered by STEM undergraduate curricula. To address them, Prof. Magana and her team have tested a set of pedagogical practices and teaching strategies that can lead to improvements in student knowledge.
 
Prof. Bedřich Beneš holds a master's degree and a PhD from the Faculty of Electrical Engineering of the Czech Technical University. He developed his career as a leading expert in computer graphics at Purdue University. In his talk, he will introduce several methods for plant modeling and simulation, and also present several advances and approaches that attempt to bridge the gap between simulation data for deep learning and real-world artificial intelligence methods for agriculture.
 
An abstract of both presentations can be found below. The lectures were supported by the Fulbright Specialist Program.

Teaching and Learning With Computation, Modeling, and Simulation

by Prof. Alejandra J. Magana

The presentation provides an overview of the opportunities and challenges of introducing computational, modeling, and simulation practices within the undergraduate STEM (Science, Technology, Engineering, Mathematics) curricula. Through a series of studies and classroom implementations, we have identified specific struggles that students encountered, and to address those struggles, we have tested a set of pedagogies and learning strategies that can result in students’ computational adaptive expertise. Lessons learned from these studies have resulted in a computational cognitive apprenticeship model that can be used as a guideline to support learners in using computation meaningfully for their learning and overcoming challenges when engaged in this complex practice. 

Bridging the Sim2Real Gap for Geometric Models in AI Phenotyping

by Prof. Bedřich Beneš

Vegetation modeling has undergone unprecedented progress in the past 40 years. It has evolved from cone-like simple static shapes to high-detailed physics-enabled models with photonic-level illumination simulating plant growth with complex intrinsic and extrinsic signaling at interactive framerates. These models are currently being complemented with deep learning methods, which require enormous amounts of labeled data that are often difficult to obtain. This talk will introduce several methods for plant modeling and simulation, and it will also present several advances and practices that attempt to bridge the gap between the simulation data for deep learning and real AI methods for agriculture. In particular, we will show how large datasets can be used to train deep networks that act as humans in estimating the visual quality of real and virtual objects. We will show how carefully designed generative models can be applied to count leaves in natural plants and how this can be applied in agriculture and genomics.