arXiv subjects

Kristy Lee

Publications and source records attributed to Kristy Lee.

2 recordsLinked to original sources

arXiv2023

Navigating in the Dark -- Designing Autonomous Driving Features to Assist Old Adults with Visual Impairments

Age-related macular degeneration is a leading cause of blindness worldwide and is one of many limitations to independent driving among old adults. Highly autonomous vehicles present a prospective solution for those who are no longer capable of driving due to low vision. However, accessibility issues must be addressed to create a safe and pleasant experience for this group of users so that it allows them to maintain an appropriate level of situational awareness and a sense of control during driving. In this study, we made use of a human-centered design process consisting of five stages - empathize, define, ideate, prototype, and test. We designed a prototype to aid old adults with age-related macular degeneration to travel with a necessary level of situational awareness and remain in control while riding in a highly or fully autonomous vehicle. The final design prototype includes a voice-activated navigation system with three levels of details to bolster situational awareness, a 360 degree in-vehicle camera to detect both the passenger and objects around the vehicle, a retractable microphone for the passenger to be easily registered in the vehicle while speaking, and a physical button on the console-side of the right and left front seats to manually activate the navigation system.

cs.HC↗

arXiv2021

Adapting Language Models for Zero-shot Learning by Meta-tuning on Dataset and Prompt Collections

Large pre-trained language models (LMs) such as GPT-3 have acquired a surprising ability to perform zero-shot learning. For example, to classify sentiment without any training examples, we can "prompt" the LM with the review and the label description "Does the user like this movie?", and ask whether the next word is "yes" or "no". However, the next word prediction training objective is still misaligned with the target zero-shot learning objective. To address this weakness, we propose meta-tuning, which directly optimizes the zero-shot learning objective by fine-tuning pre-trained language models on a collection of datasets. We focus on classification tasks, and construct the meta-dataset by aggregating 43 existing datasets and annotating 441 label descriptions in a question-answering (QA) format. When evaluated on unseen tasks, meta-tuned models outperform a same-sized QA model and the previous SOTA zero-shot learning system based on natural language inference. Additionally, increasing parameter count from 220M to 770M improves AUC-ROC scores by 6.3%, and we forecast that even larger models would perform better. Therefore, measuring zero-shot learning performance on language models out-of-the-box might underestimate their true potential, and community-wide efforts on aggregating datasets and unifying their formats can help build models that answer prompts better.

cs.CL↗