Efficient Probabilistic Inference Algorithms for Cooperative Multiagent Systems
Bibliographic record
Abstract
Probabilistic reasoning methods, Bayesian networks (BNs) in particular, have emerged as an effective and central tool for reasoning under uncertainty. In a multi-agent environment, agents equipped with local knowledge often need to collaborate and reason about a larger uncertainty domain. Multiply sectioned Bayesian networks (MSBNs) provide a solution for the probabilistic reasoning of cooperative agents in such a setting. In this thesis, we first aim to improve the efficiency of current MSBN exact inference algorithms. We show that by exploiting the calculation schema and the semantic meaning of inter-agent messages, we can significantly reduce an agent's local computational cost as well as the inter-agent communication overhead. Our novel technical contributions include 1) a new message passing architecture based on an MSBN linked junction tree forest (LJF); 2) a suite of algorithms extended from our work in BNs to provide the semantic analysis of inter-agent messages; 3) a fast marginal calibration algorithm, designed for an LJF that guarantees exact results with a minimum local and global cost. We then investigate how to incorporate approximation techniques in the MSBN framework. We present a novel local adaptive importance sampler (LLAIS) designed to apply localized stochastic sampling while maintaining the LJF structure. The LLAIS sampler provides accurate estimations for local posterior beliefs and promotes efficient calculation of inter-agent messages. We also address the problem of online monitoring for cooperative agents. As the MSBN model is restricted to static domains, we introduce an MA-DBN model based on a combination of the MSBN and dynamic Bayesian network (DBN) models. We show that effective multi-agent online monitoring with bounded error is possible in an MA-DBN through a new secondary inference structure and a factorized representation of forward messages.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".