Accurately predicting students' future performance based on their tracked academic records in college programs is crucial for effectively carrying out necessary pedagogical interventions to ensure students' on-time graduation. Although there is a rich literature on predicting student performance in solving problems and studying courses using data-driven approaches, predicting student performance in completing college programs is much less studied and faces new challenges, mainly due to the diversity of courses selected by students and the requirement of continuous tracking and incorporation of students' evolving progresses. In this paper, we develop a novel algorithm that enables progressive prediction of students' performance by adapting ensemble learning techniques and utilizing education-specific domain knowledge. We prove its prediction performance guarantee and show its performance improvement against benchmark algorithms on a real-world student dataset from UCLA.

Automated learning and decision making systems in public-facing applications are vulnerable to malicious attacks. Examples of such systems include spam detectors, credit card fraud detectors, and network intrusion detection systems. These systems are at further risk of attack when money is directly involved, such as market forecasters or decision systems used in determining insurance or loan rates. In this paper, we consider the setting where a predictor Bob has a fixed model, and an unknown attacker Alice aims to perturb (or poison) future test instances so as to alter Bob's prediction to her benefit. We focus specifically on Bob's optimal defense actions to limit Alice's effectiveness. We define a general framework for determining Bob's optimal defense action against Alice's worst-case attack. We then demonstrate our framework by considering linear predictors, where we provide tractable methods of determining the optimal defense action. Using these methods, we perform an empirical investigation of optimal defense actions for a particular class of linear models -- autoregressive forecasters -- and find that for ten real world futures markets, the optimal defense action reduces the Bob's loss by between 78 and 97%.

Consider a situation with n agents or players, where some of the players form a coalition with a certain collective objective. Simple games are used to model systems that can decide whether coalitions are successful (winning) or not (losing). A simple game can be viewed as a monotone boolean function. The dimension of a simple game is the smallest positive integer d such that the simple game can be expressed as the intersection of d threshold functions, where each threshold function uses a threshold and n weights. Taylor and Zwicker have shown that d is bounded from above by the number of maximal losing coalitions. We present two new upper bounds both containing the Taylor-Zwicker bound as a special case. The Taylor-Zwicker bound implies an upper bound of (n choose n/2). We improve this upper bound significantly by showing constructively that d is bounded from above by the cardinality of any binary covering code with length n and covering radius 1. This result supplements a recent result where Olsen et al. showed how to construct simple games with dimension |C| for any binary constant weight SECDED code C with length n. Our result represents a major step in the attempt to close the dimensionality gap for simple games.

Policy gradient methods have been widely applied in reinforcement learning. For reasons of safety and cost, learning is often conducted using a simulator. However, learning in simulation does not traditionally utilise the opportunity to improve learning by adjusting certain environment variables - state features that are randomly determined by the environment in a physical setting but controllable in a simulator. Exploiting environment variables is crucial in domains containing significant rare events (SREs), e.g., unusual wind conditions that can crash a helicopter, which are rarely observed under random sampling but have a considerable impact on expected return. We propose off environment reinforcement learning (OFFER), which addresses such cases by simultaneously optimising the policy and a proposal distribution over environment variables. We prove that OFFER converges to a locally optimal policy and show experimentally that it learns better and faster than a policy gradient baseline.

We consider the setting of stochastic multiagent systems and formulate an automated verification framework for quantifying and reasoning about agents' trust. To capture human trust, we work with a cognitive notion of trust defined as a subjective evaluation that agent A makes about agent B's ability to complete a task, which in turn may lead to a decision by A to rely on B. We propose a probabilistic rational temporal logic PRTL*, which extends the logic PCTL* with reasoning about mental attitudes (beliefs, goals and intentions), and includes novel operators that can express concepts of social trust such as competence, disposition and dependence. The logic can express, for example, that "agent A will eventually trust agent B with probability at least p that B will be have in a way that ensures the successful completion of a given task". We study the complexity of the automated verification problem and, while the general problem is undecidable, we identify restrictions on the logic and the system that result in decidable, or even tractable, subproblems.

Reinforcement Learning (RL) is a widely used solution for sequential decision-making problems and has been used in many complex domains. However, RL algorithms suffer from scalability issues, especially when multiple agents are acting in a shared environment. This research intends to accelerate learning in multiagent sequential decision-making tasks by reusing previous knowledge, both from past solutions and advising between agents. We intend to contribute a Transfer Learning framework focused on Multiagent RL, requiring as few domain-specific hand-coded parameters as possible.

We propose Kont, a formal framework for comparing normative multiagent systems (nMASs) by computing tradeoffs among liveness (something good happens) and safety (nothing bad happens). Safety-focused nMASs restrict agents' actions to avoid undesired enactments. However, such restrictions hinder liveness, particularly in situations such as medical emergencies. We formalize tradeoffs using norms, and develop an approach for understanding to what extent an nMAS promotes liveness or safety. We propose patterns to guide the design of an nMAS with respect to liveness and safety, and prove their correctness. We further quantify liveness and safety using heuristic metrics for an emergency healthcare application. We show that the results of the application corroborate our theoretical development.

The Cooperative Target Observation (CTO) problem has been of great interest in the multi-agents and robotics literature due to the problem being at the core of a number of applications including surveillance. In CTO problem, the observer agents attempt to maximize the collective time during which each moving target is being observed by at least one observer in the area of interest. However, most of the prior works for the CTO problem consider the targets movement to be Randomized. Given our focus on surveillance domain, we modify this assumption to make the targets strategic and present two target strategies namely Straight-line strategy and Controlled Randomization strategy. We then modify the observer strategy proposed in the literature based on the K-means algorithm to introduce five variants and provide experimental validation. In surveillance domain, it is often reasonable to assume that the observers may themselves be a subject of observation for a variety of purposes by unknown adversaries whose model may not be known. Randomizing the observers actions can help to make their target observation strategy less predictable. As the fifth variant, we therefore introduce Adjustable Randomization into the best performing observer strategy where the observer can adjust the expected loss in reward due to randomization depending on the situation.

In this paper, we analyze an emerging economic form, called fans economy, in which a fan donates money to the host and gets allocated proportional to the amount of his donation (normalized by the overall amount of donation). Fans economy is the major way live streaming apps monetize and includes a number of popular economic forms ranging from crowdfunding to mutual fund. We propose an auction game, coined all-pay auctions with proportional allocation (APAPA), to model the fans economy and analyze the auction from the perspective of revenue. Comparing to the standard all-pay auction, which normally has no pure Nash-Equilibrium in the complete information setting, we solve the pure Nash-Equilibrium of the APAPA in closed form and prove its uniqueness. Motivated by practical concerns, we then analyze the case where APAPA is equipped with a reserve and show that there might be multiple equilibria in this case. We give an efficient algorithm to compute all equilibria in this case. For either case, with or without reserve, we show that APAPA always extracts revenue that 2-approximates the second-highest valuation. Furthermore, we conduct experiments to show how revenue changes with respect to different reserves.

This paper studies the benefit for information providers in free public information disclosure in settings where the prospective information buyers are people. The underlying model, which applies to numerous real-life situations, considers a standard decision making setting where the decision maker is uncertain about the outcomes of her decision. The information provider can fully disambiguate this uncertainty and wish to maximize her profit from selling such information. We use a series of AMT-based experiments with people to test the benefit for the information provider from reducing some of the uncertainty associated with the decision maker's problem, for free. Free information disclosure of this kind can be proved to be ineffective when the buyer is a fully-rational agent. Yet, when it comes to people we manage to demonstrate that a substantial improvement in the information provider's profit can be achieved with such an approach. The analysis of the results reveals that the primary reason for this phenomena is people's failure to consider the strategic nature of the interaction with the information provider. Peoples' inability to properly calculate the value of information is found to be secondary in its influence.