Myra B. Cohen

Search-Based Software Testing relies on fitness functions to guide search algorithms toward target branches. Traditionally, computing fitness for branches involves a program analysis pipeline, with instrumentation that is language and type dependent. Large Language Models (LLMs) provide a promising mechanism for automatically generating branch aware executable fitness functions. However, directly using LLM generated outputs is unreliable, leading to incomplete outputs, compilation errors, or incorrect optimization objectives. This paper presents FitGen, a validation guided repair framework combining structured prompt design with output, syntax, and semantic validation to automatically diagnose and repair executable fitness functions. We evaluate FitGen on 11 benchmark programs containing 51 target branches using fitness functions generated by six state of the art LLMs. Our results show FitGen improves branch validation performance across all LLMs by up to 40.91% using the repaired fitness functions, and search based optimization achieves up to 93.59% branch validation accuracy.

Microbial carbon utilisation shapes ecological niches, community assembly, and nutrient cycling, yet remains difficult to predict from genomes despite well-characterised pathways. Genome-based machine-learning models often generalise poorly across datasets, a limitation commonly attributed to small or taxonomically biased training sets. We integrated binary growth phenotypes for 819 strains across 240 carbon sources from four datasets, focusing on 15 shared sources. Cross-dataset evaluation reduced balanced accuracy from 0.86 (within datasets) to 0.62, and restricting evaluation to in-clade relatives recovered only 0.03 of this loss. Because phylogeny alone could not explain the transfer gap, we restricted training to samples with coherent genotype-phenotype relationships, defined by agreement between experimental labels and annotated pathways. Concordance filtering doubled the number of carbon sources with reproducible genomic signals across datasets (from 6 to 12 of 15), often recovering known pathway genes, while matched random subsets remained at the unfiltered level. The smaller concordant set also approached the cross-dataset performance of models augmented with up to 1800 BacDive genomes per phenotype, demonstrating greater efficiency per sample. Concordance-trained models remained bounded specialists, but separated growers from non-growers among genomes the pathway rules called negative, mostly generalists with broad-specificity transporters. Model confidence made these bounds actionable, since abstention improved accuracy where models transferred and retraining on newly labelled low-confidence genomes gave the largest gains for weaker phenotypes. Collectively, these findings recast generalisation as a problem of label-mechanism agreement and demonstrate how genome-based predictions can be deployed within explicit applicability domains and improved through targeted phenotyping.

Small uncrewed aerial systems (sUAS) provide an invaluable resource for performing a variety of surveillance, search, and delivery tasks in remote or hostile terrains which may not be accessible by other means. Due to the critical role sUAS play in these situations, it is vital that they are well configured in order to ensure a safe and stable flight. However, it is not uncommon for mistakes to occur in configuration and calibration, leading to failures or incomplete missions. To address this problem, we propose a set of self-adaptive mechanisms and implement them into a self-adaptive framework, Controller Instability-preventing Configuration-Aware Drone Adaptation (CICADA). CICADA dynamically detects unstable drone behavior during flight and adapts to mitigate this threat. We have built a prototype of CICADA using a popular open source sUAS flight control software and experimented with a large number of different configurations in simulation. We then performed a case study with physical drones to determine if our framework will work in practice. Experimental results show that CICADA ’s adaptations reduce controller instability and enable the sUAS to recover from up to 33.8% of poor configurations. In cases where we cannot complete the intended mission, invoking alternative adaptations may still help by allowing the vehicle to loiter or land safely in place, avoiding potentially catastrophic crashes. These safety-focused adaptations can mitigate unsafe behavior in 52.9% to 64.7% of dangerous configurations. We further show that rule-based approaches can be leveraged to automatically select an appropriate adaptation strategy based on the severity of instability encountered, with up to a 14.2% improvement over direct adaptation. Finally, we introduce a variation of our primary adaptation strategy designed to allow more cautious adaptation with limited configuration information, which gets within 6.7% of our primary adaptation strategy despite not requiring an optimal knowledge base.