Adaptively forecasting human behavior in social settings is an important step toward achieving Artificial General Intelligence. Most existing research in social forecasting has focused either on unfocused interactions, such as pedestrian trajectory prediction, or on monadic and dyadic behavior forecasting. In contrast, social psychology emphasizes the importance of group interactions for understanding complex social dynamics. This creates a gap that we address in this paper: forecasting social interactions at the group (conversation) level. Additionally, it is important for a forecasting model to be able to adapt to groups unseen at train time, as even the same individual behaves differently across different groups. This highlights the need for a forecasting model to explicitly account for each group's unique dynamics. To achieve this, we adopt a meta-learning approach to human behavior forecasting, treating every group as a separate meta-learning task. As a result, our method conditions its predictions on the specific behaviors within the group, leading to generalization to unseen groups. Specifically, we introduce Social Process (SP) models, which predict a distribution over future multimodal cues jointly for all group members based on their preceding low-level multimodal cues, while incorporating other past sequences of the same group's interactions. In this work we also analyze the generalization capabilities of SP models in both their outputs and latent spaces through the use of realistic synthetic datasets.

In 1969, Tony Hoare published a classical Communications' article, "An Axiomatic Basis for Computer Programming." Hoare's article culminated a sequence of works by Turing, McCarthy, Wirth, Floyd, and Manna, whose essence is an association of a proposition with each point in the program control flow, where the proposition is asserted to hold whenever that point is reach. Hoare added two important elements to that approach. First, he described a formal logic, now called Hoare Logic, for reasoning about programs. Second, he offered a compelling vision for the program-verification project: "When the correctness of a program, its compiler, and the hardware of the computer have all been established with mathematical certainty, it will be possible to place great reliance on the results of the program, and predict their properties with a confidence limited only by the reliability of the electronics."

Computational models of aggregated social agents have two major faults: (1) inter-individual entrainment is ignored; and (2) rule-sets governing behavior are invariant to history. Together these shortcomings impede our ability to generate realistic models of complex evolving social processes. To illustrate how even simple couplings within an established dyad generates unexpected outcomes, we present our findings from two computer models (agent-based, particle filter) of married couples. With the use of computational modeling, especially when attempting to capture and articulate trajectories of socially aggregated agents, numerous implicit assumptions are made and yet, many if not most, are without an empirical Figure 1: User interface showing parameter sliderbars that foundation. For example, the standard protocol for creating modify interaction characteristics.