Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userI Dont Think So: Disagreement-Based Policy Summaries for Comparing Agents
107
0
0.0
(
0
)
Added by
Yotam Amitai
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
With Artificial Intelligence on the rise, human interaction with autonomous agents becomes more frequent. Effective human-agent collaboration requires that the human understands the agents behavior, as failing to do so may lead to reduced productiveness, misuse, frustration and even danger. Agent strategy summarization methods are used to describe the strategy of an agent to its destined user through demonstration. The summarys purpose is to maximize the users understanding of the agents aptitude by showcasing its behaviour in a set of world states, chosen by some importance criteria. While shown to be useful, we show that these methods are limited in supporting the task of comparing agent behavior, as they independently generate a summary for each agent. In this paper, we propose a novel method for generating contrastive summaries that highlight the differences between agents policies by identifying and ranking states in which the agents disagree on the best course of action. We conduct a user study in which participants face an agent selection task. Our results show that the novel disagreement-based summaries lead to improved user performance compared to summaries generated using HIGHLIGHTS, a previous strategy summarization algorithm.
rate research
Read More
By studying the underlying policies of decision-making agents, we can learn about their shortcomings and potentially improve them. Traditionally, this has been done either by examining the agents implementation, its behaviour while it is being executed, its performance with a reward/fitness function or by visualizing the density of states the agent visits. However, these methods fail to describe the policys behaviour in complex, high-dimensional environments or do not scale to thousands of policies, which is required when studying training algorithms. We propose policy supervectors for characterizing agents by the distribution of states they visit, adopting successful techniques from the area of speech technology. Policy supervectors can characterize policies regardless of their design philosophy (e.g. rule-based vs. neural networks) and scale to thousands of policies on a single workstation machine. We demonstrate methods applicability by studying the evolution of policies during reinforcement learning, evolutionary training and imitation learning, providing insight on e.g. how the search space of evolutionary algorithms is also reflected in agents behaviour, not just in the parameters.
Research in developmental psychology consistently shows that children explore the world thoroughly and efficiently and that this exploration allows them to learn. In turn, this early learning supports more robust generalization and intelligent behavior later in life. While much work has gone into developing methods for exploration in machine learning, artificial agents have not yet reached the high standard set by their human counterparts. In this work we propose using DeepMind Lab (Beattie et al., 2016) as a platform to directly compare child and agent behaviors and to develop new exploration techniques. We outline two ongoing experiments to demonstrate the effectiveness of a direct comparison, and outline a number of open research questions that we believe can be tested using this methodology.
Tree Search (TS) is crucial to some of the most influential successes in reinforcement learning. Here, we tackle two major challenges with TS that limit its usability: textit{distribution shift} and textit{scalability}. We first discover and analyze a counter-intuitive phenomenon: action selection through TS and a pre-trained value function often leads to lower performance compared to the original pre-trained agent, even when having access to the exact state and reward in future steps. We show this is due to a distribution shift to areas where value estimates are highly inaccurate and analyze this effect using Extreme Value theory. To overcome this problem, we introduce a novel off-policy correction term that accounts for the mismatch between the pre-trained value and its corresponding TS policy by penalizing under-sampled trajectories. We prove that our correction eliminates the above mismatch and bound the probability of sub-optimal action selection. Our correction significantly improves pre-trained Rainbow agents without any further training, often more than doubling their scores on Atari games. Next, we address the scalability issue given by the computational complexity of exhaustive TS that scales exponentially with the tree depth. We introduce Batch-BFS: a GPU breadth-first search that advances all nodes in each depth of the tree simultaneously. Batch-BFS reduces runtime by two orders of magnitude and, beyond inference, enables also training with TS of depths that were not feasible before. We train DQN agents from scratch using TS and show improvement in several Atari games compared to both the original DQN and the more advanced Rainbow.
Fairness in algorithmic decision-making processes is attracting increasing concern. When an algorithm is applied to human-related decision-making an estimator solely optimizing its predictive power can learn biases on the existing data, which motivates us the notion of fairness in machine learning. while several different notions are studied in the literature, little studies are done on how these notions affect the individuals. We demonstrate such a comparison between several policies induced by well-known fairness criteria, including the color-blind (CB), the demographic parity (DP), and the equalized odds (EO). We show that the EO is the only criterion among them that removes group-level disparity. Empirical studies on the social welfare and disparity of these policies are conducted.
Maintaining the stability of the modern power grid is becoming increasingly difficult due to fluctuating power consumption, unstable power supply coming from renewable energies, and unpredictable accidents such as man-made and natural disasters. As the operation on the power grid must consider its impact on future stability, reinforcement learning (RL) has been employed to provide sequential decision-making in power grid management. However, existing methods have not considered the environmental constraints. As a result, the learned policy has risk of selecting actions that violate the constraints in emergencies, which will escalate the issue of overloaded power lines and lead to large-scale blackouts. In this work, we propose a novel method for this problem, which builds on top of the search-based planning algorithm. At the planning stage, the search space is limited to the action set produced by the policy. The selected action strictly follows the constraints by testing its outcome with the simulation function provided by the system. At the learning stage, to address the problem that gradients cannot be propagated to the policy, we introduce Evolutionary Strategies (ES) with black-box policy optimization to improve the policy directly, maximizing the returns of the long run. In NeurIPS 2020 Learning to Run Power Network (L2RPN) competition, our solution safely managed the power grid and ranked first in both tracks.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
