• Contact Info
  • +972-528105556
Publications in VIVO

Shani, Guy


Guy Shani is active in the field of automated planning under partial observability. He developed several competitive solvers for POMDPs, mainly the FSVI algorithm. In addition, Guy has co-authored the survey on point-based POMDP solvers, which provides both an in depth introduction to point-based solvers, as well as a comprehensive empirical comparison of the various components of point-based solvers.

Later on, Guy has shifted his interests into the area of qualitative models for partial observability. He has contributed much to the development of scalable algorithms for online contingent planning under partial observability. The SDR, MPSR, and HCP algorithms are perhaps the most scalable solvers to date, managing to handle domains much larger than previous state-of-the-art solvers.  Guy has helped developing the CPOR solver, which is currently the most scalable solver for producing complete plan graphs for partially observable contingent problems. Guy has also helped to construct the QDec-POMDP model – a qualitative alternative for Dec-PMDP models for multi agent collaborative planning under partial observability.

In the last few years Guy has focused on planning with privacy constraints. Guy took part in the development of a wide range of planning approaches for privacy preserving plans (PPP). His work on adapting heuristics, such as landmarks and pattern databases, for PPP was key in the development of PPP algorithms. Guy also helped producing stronger projections that better capture agent dependencies.  Guy also worked on a method, based on action macros, that reduces the dependencies between non-neighboring agents.

Aside for planning, Guy is also active in the field of recommender systems. His work on evaluating recommendation systems in the recommender systems handbook has received much attention in that community. He has also explored many other problems in recommender systems, such as using social networks, Wikipedia, and web resources to collect information for recommendations, and many more.

Finally, Guy is also active in applied machine learning in agriculture, and has published papers that use CNN deep learning models for identifying various plant attributes, such as disease symptoms in potatoes, corn and wheat stems, tracking tomato flowers in greenhouses, and more.

Research Areas research areas

selected publications