Damian Dailisan

Marcin Korecki, Damian Dailisan, Cesare Carissimo · eth-zurich

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 with are dynamically changing, using a multiple-objective approach. We apply this approach to extend Deep $Q$-Learning to accommodate multiple objectives and evaluate this method on a simplified two-leg intersection controlled by a switching agent.Our approach dynamically accommodates the preferences of drivers on the system and achieves better overall performance across three metrics (speeds, stops, and waits) while integrating objectives that have competing or conflicting actions.

Ritz Ann Aguilar, Damian Dailisan · eth-zurich

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 basis patterns when used for SPI is limited. We present a modified deep convolutional autoencoder network (DCAN) for SPI on 64x64 pixel images with up to 6.25% compression ratio and apply binary and orthogonality regularizers during training. Training a DCAN with these regularizers allows it to learn multiple measurement bases that have combinations of binary or non-binary, and orthogonal or non-orthogonal patterns. We compare the reconstruction quality, orthogonality of the patterns, and robustness to noise of the resulting DCAN models to traditional SPI reconstruction algorithms (such as Total Variation minimization and Fourier Transform). Our DCAN models can be trained to be robust to noise while still having fast enough reconstruction times (~3 ms per frame) to be viable for real-time imaging.

Joshua C. Yang, Damian Dailisan, Marcin Korecki et al. · eth-zurich

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 observed that the choice of voting methods and the presentation order influenced LLM voting outcomes. We found that varying the persona can reduce some of these biases and enhance alignment with human choices. While the Chain-of-Thought approach did not improve prediction accuracy, it has potential for AI explainability in the voting process. We also identified a trade-off between preference diversity and alignment accuracy in LLMs, influenced by different temperature settings. Our findings indicate that LLMs may lead to less diverse collective outcomes and biased assumptions when used in voting scenarios, emphasizing the need for cautious integration of LLMs into democratic processes.