No description available.

Text-to-SQL, which involves translating natural language into Structured Query Language (SQL), is crucial for enabling broad access to structured databases without expert knowledge. However, designing models for such tasks is challenging due to numerous factors, including the presence of ’noise,’ such as ambiguous questions and syntactical errors. This thesis provides an in-depth analysis of the distribution and types of noise in the widely used BIRD-Bench benchmark and the impact of noise on models. While BIRD-Bench was created to model dirty and noisy database values, it was not created to contain noise and errors in the questions and gold queries. We found after a manual evaluation that noise in questions and gold queries are highly prevalent in the financial domain of the dataset, and a further analysis of the other domains indicate the presence of noise in other parts as well. The presence of incorrect gold SQL queries, which then generate incorrect gold answers, has a significant impact on the benchmark’s reliability. Surprisingly, when evaluating models on corrected SQL queries, zero-shot baselines surpassed the performance of state-of-the-art prompting methods. The thesis then introduces the concept of classifying noise in natural language questions, aiming to prevent the entry of noisy questions into text-to-SQL models and to annotate noise in existing datasets. Experiments using GPT-3.5 and GPT-4 on a manually annotated dataset demonstrated the viability of this approach, with classifiers achieving up to 0.81 recall and 80% accuracy. Additionally, the thesis explored the use of LLMs for automatically correcting faulty SQL queries. This showed a 100% success rate for specific query corrections, highlighting the potential for LLMs in improving dataset quality. We conclude that informative noise labels and reliable benchmarks are crucial to developing new Text-to-SQL methods that can handle varying types of noise.

Life on Earth is strongly related to the Sun, which makes it a vital star to study and understand. To improve our knowledge of the way the Sun works, many satellites have been launched into space to monitor the Sun‟s activities where the one of main focus is the effect of these activities on the Earth‟s climate; PICARD is one such satellite. Due to the noise associated with SODISM images, the clarity of these images and the appearance of solar features are affected. Image denoising and enhancement are the main techniques to improve the visual appearance of SODISM images. Affective de-noising algorithm methods depend on a proper detecting of noise present in the image. The aim is to identify which type of noise is present in the image. To reach this point, supervised machine-learning (ML) classifier is used to classify the type of noise present in the image. Furthermore, this work introduces a novel technique developed to enhance the quality of SODISM images. In this thesis, the Modified Undecimated Discrete Wavelet Transform (M-UDWT) technique is used to de-noise and enhance the quality of SODISM images. The proposed method is robust and effectively improves the quality of SODISM images, and produces more precise information and clear feature are brought out. In addition, the non wavelet enhancement is developed as well in this thesis. The results of this algorithm is discussed. The new methods are also assessed using two different methods: subjective (by human observation) and objective (by calculation)