arXiv subjects

Yigit Gurses

Publications and source records attributed to Yigit Gurses.

4 recordsLinked to original sources

arXiv2025

Am I Blue or Is My Hobby Counting Teardrops? Expression Leakage in Large Language Models as a Symptom of Irrelevancy Disruption

Large language models (LLMs) have advanced natural language processing (NLP) skills such as through next-token prediction and self-attention, but their ability to integrate broad context also makes them prone to incorporating irrelevant information. Prior work has focused on semantic leakage, bias introduced by semantically irrelevant context. In this paper, we introduce expression leakage, a novel phenomenon where LLMs systematically generate sentimentally charged expressions that are semantically unrelated to the input context. To analyse the expression leakage, we collect a benchmark dataset along with a scheme to automatically generate a dataset from free-form text from common-crawl. In addition, we propose an automatic evaluation pipeline that correlates well with human judgment, which accelerates the benchmarking by decoupling from the need of annotation for each analysed model. Our experiments show that, as the model scales in the parameter space, the expression leakage reduces within the same LLM family. On the other hand, we demonstrate that expression leakage mitigation requires specific care during the model building process, and cannot be mitigated by prompting. In addition, our experiments indicate that, when negative sentiment is injected in the prompt, it disrupts the generation process more than the positive sentiment, causing a higher expression leakage rate.

cs.CL↗

arXiv2023

GRJointNET: Synergistic Completion and Part Segmentation on 3D Incomplete Point Clouds

Segmentation of three-dimensional (3D) point clouds is an important task for autonomous systems. However, success of segmentation algorithms depends greatly on the quality of the underlying point clouds (resolution, completeness etc.). In particular, incomplete point clouds might reduce a downstream model's performance. GRNet is proposed as a novel and recent deep learning solution to complete point clouds, but it is not capable of part segmentation. On the other hand, our proposed solution, GRJointNet, is an architecture that can perform joint completion and segmentation on point clouds as a successor of GRNet. Features extracted for the two tasks are also utilized by each other to increase the overall performance. We evaluated our proposed network on the ShapeNet-Part dataset and compared its performance to GRNet. Our results demonstrate GRJointNet can outperform GRNet on point completion. It should also be noted that GRNet is not capable of segmentation while GRJointNet is. This study1, therefore, holds a promise to enhance practicality and utility of point clouds in 3D vision for autonomous systems.

cs.CV↗

arXiv2023

Developing Driving Strategies Efficiently: A Skill-Based Hierarchical Reinforcement Learning Approach

Driving in dense traffic with human and autonomous drivers is a challenging task that requires high-level planning and reasoning. Human drivers can achieve this task comfortably, and there has been many efforts to model human driver strategies. These strategies can be used as inspirations for developing autonomous driving algorithms or to create high-fidelity simulators. Reinforcement learning is a common tool to model driver policies, but conventional training of these models can be computationally expensive and time-consuming. To address this issue, in this paper, we propose ``skill-based" hierarchical driving strategies, where motion primitives, i.e. skills, are designed and used as high-level actions. This reduces the training time for applications that require multiple models with varying behavior. Simulation results in a merging scenario demonstrate that the proposed approach yields driver models that achieve higher performance with less training compared to baseline reinforcement learning methods.

cs.LG↗

arXiv2023

Neural Network Adaptive Control with Long Short-Term Memory

In this study, we propose a novel adaptive control architecture, which provides dramatically better transient response performance compared to conventional adaptive control methods. What makes this architecture unique is the synergistic employment of a traditional, Adaptive Neural Network (ANN) controller and a Long Short-Term Memory (LSTM) network. LSTM structures, unlike the standard feed-forward neural networks, can take advantage of the dependencies in an input sequence, which can contain critical information that can help predict uncertainty. Through a novel training method we introduced, the LSTM network learns to compensate for the deficiencies of the ANN controller during sudden changes in plant dynamics. This substantially improves the transient response of the system and allows the controller to quickly react to unexpected events. Through careful simulation studies, we demonstrate that this architecture can improve the estimation accuracy on a diverse set of uncertainties for an indefinite time span. We also provide an analysis of the contributions of the ANN controller and LSTM network to the control input, identifying their individual roles in compensating low and high-frequency error dynamics. This analysis provides insight into why and how the LSTM augmentation improves the system's transient response. The stability of the overall system is also shown via a rigorous Lyapunov analysis.

eess.SY↗