Paul B. Kantor

Eric Pulick, Shubham Bharti, Yiding Chen et al.

As machine learning (ML) is more tightly woven into society, it is imperative that we better characterize ML's strengths and limitations if we are to employ it responsibly. Existing benchmark environments for ML, such as board and video games, offer well-defined benchmarks for progress, but constituent tasks are often complex, and it is frequently unclear how task characteristics contribute to overall difficulty for the machine learner. Likewise, without a systematic assessment of how task characteristics influence difficulty, it is challenging to draw meaningful connections between performance in different benchmark environments. We introduce a novel benchmark environment that offers an enormous range of ML challenges and enables precise examination of how task elements influence practical difficulty. The tool frames learning tasks as a "board-clearing game," which we call the Game of Hidden Rules (GOHR). The environment comprises an expressive rule language and a captive server environment that can be installed locally. We propose a set of benchmark rule-learning tasks and plan to support a performance leader-board for researchers interested in attempting to learn our rules. GOHR complements existing environments by allowing fine, controlled modifications to tasks, enabling experimenters to better understand how each facet of a given learning task contributes to its practical difficulty for an arbitrary ML algorithm.

Eric Pulick, Vladimir Menkov, Yonatan Mintz et al.

Reliable real-world deployment of reinforcement learning (RL) methods requires a nuanced understanding of their strengths and weaknesses and how they compare to those of humans. Human-machine systems are becoming more prevalent and the design of these systems relies on a task-oriented understanding of both human learning (HL) and RL. Thus, an important line of research is characterizing how the structure of a learning task affects learning performance. While increasingly complex benchmark environments have led to improved RL capabilities, such environments are difficult to use for the dedicated study of task structure. To address this challenge we present a learning environment built to support rigorous study of the impact of task structure on HL and RL. We demonstrate the environment's utility for such study through example experiments in task structure that show performance differences between humans and RL algorithms.

Christo Mathew, Wentian Wang, Jacob Feldman et al.

We investigate reinforcement learning in the Game Of Hidden Rules (GOHR) environment, a complex puzzle in which an agent must infer and execute hidden rules to clear a 6$\times$6 board by placing game pieces into buckets. We explore two state representation strategies, namely Feature-Centric (FC) and Object-Centric (OC), and employ a Transformer-based Advantage Actor-Critic (A2C) algorithm for training. The agent has access only to partial observations and must simultaneously infer the governing rule and learn the optimal policy through experience. We evaluate our models across multiple rule-based and trial-list-based experimental setups, analyzing transfer effects and the impact of representation on learning efficiency.

Wentian Wang, Sarthak Jain, Paul Kantor et al.

We propose MMLU-SR, a novel dataset designed to measure the true comprehension abilities of Large Language Models (LLMs) by challenging their performance in question-answering tasks with modified terms. We reasoned that an agent that "truly" understands a concept can still evaluate it when key terms are replaced by suitably defined alternate terms, and sought to differentiate such comprehension from mere text replacement. In our study, we modified standardized test questions by replacing a key term with a dummy word along with its definition. The key term could be in the context of questions, answers, or both questions and answers. Notwithstanding the high scores achieved by recent popular LLMs on the MMLU leaderboard, we found a substantial reduction in model performance after such replacement, suggesting poor comprehension. This new benchmark provides a rigorous benchmark for testing true model comprehension, and poses a challenge to the broader scientific community.

Paul B. Kantor, Dennis E. Egan, Jonathan Bullinger et al.

Information sharing is vital in resisting cyberattacks, and the volume and severity of these attacks is increasing very rapidly. Therefore responders must triage incoming warnings in deciding how to act. This study asked a very specific question: "how can the addition of confidence information to alerts and warnings improve overall resistance to cyberattacks." We sought, in particular, to identify current practices, and if possible, to identify some "best practices." The research involved literature review and interviews with subject matter experts at every level from system administrators to persons who develop broad principles of policy. An innovative Modified Online Delphi Panel technique was used to elicit judgments and recommendations from experts who were able to speak with each other and vote anonymously to rank proposed practices.

Vicki Bier, Paul B. Kantor, Gary Lupyan et al.

What makes a task relatively more or less difficult for a machine compared to a human? Much AI/ML research has focused on expanding the range of tasks that machines can do, with a focus on whether machines can beat humans. Allowing for differences in scale, we can seek interesting (anomalous) pairs of tasks T, T'. We define interesting in this way: The "harder to learn" relation is reversed when comparing human intelligence (HI) to AI. While humans seems to be able to understand problems by formulating rules, ML using neural networks does not rely on constructing rules. We discuss a novel approach where the challenge is to "perform well under rules that have been created by human beings." We suggest that this provides a rigorous and precise pathway for understanding the difference between the two kinds of learning. Specifically, we suggest a large and extensible class of learning tasks, formulated as learning under rules. With these tasks, both the AI and HI will be studied with rigor and precision. The immediate goal is to find interesting groundtruth rule pairs. In the long term, the goal will be to understand, in a generalizable way, what distinguishes interesting pairs from ordinary pairs, and to define saliency behind interesting pairs. This may open new ways of thinking about AI, and provide unexpected insights into human learning.

Paul B. Kantor

We consider the problem of eliciting expert assessments of an uncertain parameter. The context is risk control, where there are, in fact, three uncertain parameters to be estimates. Two of these are probabilities, requiring the that the experts be guided in the concept of "uncertainty about uncertainty." We propose a novel formulation for expert estimates, which relies on the range and the median, rather than the variance and the mean. We discuss the process of elicitation, and provide precise formulas for these new distributions.

Vladimir Menkov, Paul Kantor

There are many legacy databases, and related stores of information that are maintained by distinct organizations, and there are other organizations that would like to be able to access and use those disparate sources. Among the examples of current interest are such things as emergency room records, of interest in tracking and interdicting illicit drugs, or social media public posts that indicate preparation and intention for a mass shooting incident. In most cases, this information is discovered too late to be useful. While agencies responsible for coordination are aware of the potential value of contemporaneous access to new data, the costs of establishing a connection are prohibitive. The problem grown even worse with the proliferation of ``hash-tagging,'' which permits new labels and ontological relations to spring up overnight. While research interest has waned, the need for powerful and inexpensive tools enabling prompt access to multiple sources has grown ever more pressing. This paper describes techniques for computing alignment matrix coefficients, which relate the fields or content of one database to those of another, using the Bayesian Ontology Alignment tool (BOA). Particular attention is given to formulas that have an easy-to-understand meaning when all cells of the data sources containing values from some small set. These formulas can be expressed in terms of probability estimates. The estimates themselves are given by a ``black box'' polytomous logistic regression model (PLRM), and thus can be easily generalized to the case of any arbitrary probability-generating model. The specific PLRM model used in this example is the BOXER Bayesian Extensible Online Regression model.