More Like this

1. State Transition in Multi-agent Epistemic Domains Using Answer Set Programming

   Izmirlioglu, Yusuf; Pham, Loc; Cao Son, Tran; Pontelli, Enrico (August 2022, Lecture notes in computer science)

   In this paper we develop a state transition function for partially observable multi-agent epistemic domains and implement it using Answer Set Programming (ASP). The transition function computes the next state upon an occurrence of a single action. Thus it can be used as a module in epistemic planners. Our transition function incorporates ontic, sensing and announcement actions and allows for arbitrary nested belief formulae and general common knowledge. A novel feature of our model is that upon an action occurrence, an observing agent corrects his (possibly wrong) initial beliefs about action precondition and his observability. By examples, we show that this step is necessary for robust state transition. We establish some properties of our state transition function regarding its soundness in updating beliefs of agents consistent with their observability. 

   more » « less
2. Momentum for the Win: Collaborative Federated Reinforcement Learning across Heterogeneous Environments

   Wang, Han; He, Shone; Zhang, Zhili; Miao, Fei; Anderson, James (May 2024, Proceedings of Machine Learning Research)

   We explore a Federated Reinforcement Learning (FRL) problem where agents collaboratively learn a common policy without sharing their trajectory data. To date, existing FRL work has primarily focused on agents operating in the same or ``similar" environments. In contrast, our problem setup allows for arbitrarily large levels of environment heterogeneity. To obtain the optimal policy which maximizes the average performance across all potentially completely different environments, we propose two algorithms: FedSVRPG-M and FedHAPG-M. In contrast to existing results, we demonstrate that both FedSVRPG-M and FedHAPG-M, both of which leverage momentum mechanisms, can exactly converge to a stationary point of the average performance function, regardless of the magnitude of environment heterogeneity. Furthermore, by incorporating the benefits of variance-reduction techniques or Hessian approximation, both algorithms achieve state-of-the-art convergence results, characterized by a sample complexity of $$O(\epsilon^{-\frac{3}{2}}/N)$$. Notably, our algorithms enjoy linear convergence speedups with respect to the number of agents, highlighting the benefit of collaboration among agents in finding a common policy. 

   more » « less
3. The Trees in the Forest: Characterizing Computing Students' Individual Help-Seeking Approaches

   https://doi.org/10.1145/3632620.3671099  

   Ko, Shao-Heng; Stephens-Martinez, Kristin (August 2024, ACM)

   Background and Context. Academic help-seeking is vital to post-secondary computing students’ effective learning. However, most empirical works in this domain study students’ help resource selection and utilization by aggregating the entire student body as a whole. Moreover, existing theoretical frameworks often implicitly assume that whether/how much a student seeks help from a specific resource only depends on context (the type of help needed and the properties of the resources), not the individual student. Objectives. To address the gap, we seek to investigate individual computing students’ help-seeking approaches by analyzing what help-seeking characteristics are individual-driven (and thus stay consistent for the same student across different course contexts) and what are context-driven. Method. We analyzed N = 597 students’ survey responses on their help resource utilization as well as their actual help-seeking records across 6 courses. We examined relations between individual students’ frequency-based help usage metrics, type-of-help requested in office/consulting hours, self-reported order of ideal help resource usage, and their collaboration inclination in small-scale sections. Findings. We found that students’ frequency-based help metrics and their order of ideal help resource usage stays relatively consistent across different course contexts, and thus may be treated as part of students’ individual help-seeking approaches. On the other hand, the type of help students seek in office/consulting hours and how much they collaborate with peers in small sections do not seem to stay consistent across different contexts and thus might be deemed more context-driven than individual-driven. Implications. Our findings reveal that part of students’ help-seeking characteristics is individual-driven. This opens up a possibility for institutions to track students’ help-seeking records in early/introductory courses, so that some preliminary understanding of students can be acquired before they enter downstream courses. Our insights may also help instructors identify which part of students’ help-seeking behavior are more likely to be influenced by their course context and design. 

   more » « less
4. A deep dive into microphone hardware for recording collaborative group work

   https://doi.org/10.5281/ZENODO.6852954  

   Bradford, Mariah; Hansen, Paige; Ross, J. Beveridge; Krishnaswamy, Nikhil; Blanchard, Nathaniel (January 2022, Educational Data Mining Conference)

   Mitrovic, Antonija; Bosch, Nigel (Ed.)

   Classroom environments are challenging for artificially intelligent agents primarily because classroom noise dilutes the interpretability and usefulness of gathered data. This problem is exacerbated when groups of students participate in collaborative problem solving (CPS). Here, we examine how well six popular microphones capture audio from individual groups. A primary usage of audio data is automatic speech recognition (ASR), therefore we evaluate our recordings by examining the accuracy of downstream ASR using the Google Cloud Platform. We simultaneously captured the audio of all microphones for 11 unique groups of three participants first reading a prepared script, and then participating in a collaborative problem solving exercise. We vary participants, noise conditions, and speech contexts. Transcribed speech was evaluated using word error rate (WER). We find that scripted speech is transcribed with a surprisingly high degree of accuracy across groups (average WER = 0.114, SD = 0.044). However, the CPS task was much more difficult (average WER = 0.570, SD = 0.143). We found most microphones were robust to background noise below a certain threshold, but the AT-Cardioid and ProCon microphones were more robust to higher noise levels. Finally, an analysis of errors revealed that most errors were due to the ASR missing words/phrases, rather than mistranscribing them. We conclude with recommendations based on our observations. 

   more » « less
5. Computational exploration of treadmilling and protrusion growth observed in fire ant rafts

   https://doi.org/10.1371/journal.pcbi.1009869  

   Wagner, Robert J.; Vernerey, Franck J. (February 2022, PLOS Computational Biology)

   Maini, Philip K. (Ed.)

   Collective living systems regularly achieve cooperative emergent functions that individual organisms could not accomplish alone. The rafts of fire ants (Solenopsis invicta) are often studied in this context for their ability to create aggregated structures comprised entirely of their own bodies, including tether-like protrusions that facilitate exploration of and escape from flooded environments. While similar protrusions are observed in cytoskeletons and cellular aggregates, they are generally dependent on morphogens or external gradients leaving the isolated role of local interactions poorly understood. Here we demonstrate through an ant-inspired, agent-based numerical model how protrusions in ant rafts may emerge spontaneously due to local interactions. The model is comprised of a condensed structural network of agents that represents the monolayer of interconnected worker ants, which floats on the water and gives ant rafts their form. Experimentally, this layer perpetually contracts, which we capture through the pairwise contraction of all neighboring structural agents at a strain rate of d ˙ . On top of the structural layer, we model a dispersed, on-lattice layer of motile agents that represents free ants, which walk on top of the floating network. Experimentally, these self-propelled free ants walk with some mean persistence length and speed that we capture through an ant-inspired phenomenological model. Local interactions occur between neighboring free ants within some radius of detection, R , and the persistence length of freely active agents is tuned through a noise parameter, η as introduced by the Vicsek model. Both R and η where fixed to match the experimental trajectories of free ants. Treadmilling of the raft occurs as agents transition between the structural and free layers in accordance with experimental observations. Ultimately, we demonstrate how phases of exploratory protrusion growth may be induced by increased ant activity as characterized by a dimensionless parameter, A . These results provide an example in which functional morphogenesis of a living system may emerge purely from local interactions at the constituent length scale, thereby providing a source of inspiration for the development of decentralized, autonomous active matter and swarm robotics. 

   more » « less