Damian Dailisan

Verified

Damian verified their affiliation via an institutional email.

Learn more

Verified

Damian verified their affiliation via an institutional email.

Learn more

• Doctor of Philosophy
• PostDoc Position at ETH Zurich

Varied real world systems such as transportation networks, supply chains and energy grids present coordination problems where many agents must learn to share resources. It is well known that the independent and selfish interactions of agents in these systems may lead to inefficiencies, often referred to as the `Price of Anarchy'. Effective interven...

We present an ethical decision-making framework that refines a pre-trained reinforcement learning (RL) model using a task-agnostic ethical layer. Following initial training, the RL model undergoes ethical fine-tuning, where human feedback is replaced by feedback generated from a large language model (LLM). The LLM embodies consequentialist, deontol...

This paper investigates the voting behaviors of Large Language Models (LLMs), specifically GPT-4 and LLaMA-2, their biases, and how they align with human voting patterns. Our methodology involved using a dataset from a human voting experiment to establish a baseline for human preferences and conducting a corresponding experiment with LLM agents. We...

To improve the performance of systems, optimization has been the prevailing approach in the past. However, the approach faces challenges when multiple goals shall be simultaneously achieved. For illustration, we study a multi-agent system, where agents have a plurality of different, and mutually inconsistent goals. We then allow agents in the syste...

This paper investigates the voting behaviors of Large Language Models (LLMs), particularly OpenAI's GPT-4 and LLaMA-2, and their alignment with human voting patterns. Our approach included a human voting experiment to establish a baseline for human preferences and a parallel experiment with LLM agents. The study focused on both collective outcomes...

Traffic on roads, packets on the Internet, and electricity on power grids share a structure abstracted in congestion games, where self-interested behaviour can lead to socially sub-optimal results. External recommendations may seek to alleviate these issues, but recommenders must take into account the effect that their recommendations have on the s...

The rapid urbanization of cities often brings about complex mobility issues, such as traffic congestion that, when unplanned, results in decreased productivity and quality of life. While many cities have adopted smart city initiatives to capture and monitor mobility, applying these in a developing country context remains a challenge when infrastruc...

The development of ethical AI systems is currently geared toward setting objective functions that align with human objectives. However, finding such functions remains a research challenge, while in RL, setting rewards by hand is a fairly standard approach. We present a methodology for dynamic value alignment, where the values that are to be aligned...

To improve the performance of systems, optimization has been the prevailing approach in the past. However, the approach faces challenges when multiple goals shall be simultaneously achieved. For illustration, we study a multi-agent system, where agents have a plurality of different, and mutually inconsistent goals. Taking decisions based on suitabl...

Traffic on roads, packets on the Internet, and electricity on power grids share a structure abstracted in congestion games, where self-interested behaviour can lead to socially sub-optimal results. External recommendations may seek to alleviate these issues, but recommenders must take into account the effect that their recommendations have on the s...

We explore the interaction of an omniscient recommender system and its users in a multiplayer routing game. Machine learning is used to model the learning dynamics of users. Recommendations provide coordination information without forcing agents to follow, but an omniscient recommender is capable of manipulating the learning dynamics. This gives ri...

Data-driven and machine-learning-based methods are increasingly used in attempts to master the challenges of the world. But are they really the best approaches to manage complex dynamical systems? Our aim is to gain more insights into this question by studying various popular reinforcement learning methods for traffic signal control, namely in disr...

Data-driven and machine-learning-based methods are increasingly used in attempts to master the challenges of the world. But are they really the best approaches to manage complex dynamical systems? Our aim is to gain more insights into this question by studying various popular reinforcement learning methods for traffic signal control, namely in disr...

Single-pixel imaging (SPI) is a novel, unconventional method that goes beyond the notion of traditional cameras but can be computationally expensive and slow for real-time applications. Deep learning has been proposed as an alternative approach for solving the SPI reconstruction problem, but a detailed analysis of its performance and generated basi...

Anticipating the increase in water demand in an urban area requires us to properly understand daily human movement driven by population size, land use, and amenity types among others. Mobility data from phones can capture human movement, but not only is this hard to obtain, but it also does not tell where the population is going. Previous studies h...

One of the most critical components of an urban transportation system is the coordination of intersections in arterial networks. With the advent of data-driven approaches for traffic control systems, deep reinforcement learning (RL) has gained significant traction in traffic control research. Proposed deep RL solutions to traffic control are design...

We modify the Nagel-Schreckenberg (NaSch) cellular automata model to study mixed-traffic dynamics. We focus on the interplay between passenger availability and bus-stopping constraints. Buses stop next to occupied cells of a discretized sidewalk model. By parametrizing the spacing distance between designated stops, our simulation covers the range o...

We modify the Nagel–Schreckenberg (NaSch) cellular automata model to study mixed-traffic dynamics. We focus on the interplay between passenger availability and bus-stopping constraints. Buses stop next to occupied cells of a discretized sidewalk model. By parametrizing the spacing distance between designated stops, our simulation covers the range o...

Given school enrollments but in the absence of a student residence census, we present a gravity-like model to infer the residential distribution of enrolled students across various administrative units. Multi-scale analysis of the effects of aggregation across different administrative levels allows for the identification of administrative units wit...

Lane changing and vehicular slowdowns are known to impact traffic flow. Using a modified Nagel–Schreckenberg cellular automata model for two vehicle types: blocking (e.g. cars) and non-blocking (e.g. motorcycles), we determined the thresholds at which the interplay of lane changing, random and non-random slowdowns strongly impact vehicle speeds. La...

Designating lanes for different vehicle types is ideal road safety-wise. Practical considerations, however, require road sharing. Using a modified Nagel–Schreckenberg cellular automata model for two vehicle types (cars and motorcycles), we analyzed the interplay of lane discipline, lane changing, and vehicle density. In the absence of lane changing...