Poo Kuan Hoong, Ph.D’s Post

Agents in Software Engineering Survey of the studies on combining LLM-based agents with Software Engineering (SE) and present a framework of LLM-based agents in SE which includes three key modules: perception, memory, and action. https://lnkd.in/g9sMKyGq

How can we harness the power of Large Language Models (LLMs) without the high computational and time costs of fine-tuning? How do we use an LLM if our resources and data are scarce? These questions have been central to our research, leading to a study I’m excited to share. Our paper, "An Analysis of LLM Fine-Tuning and Few-Shot Learning for Flaky Test Detection and Classification," has been accepted to the International Conference on Software Testing, Validation, and Verification (ICST 2025)! 🎉 In this work, we explore FlakyXbert, a Few-Shot Learning (FSL) approach that aims to reduce the resource demands of LLMs while maintaining competitive performance. Our findings examine the trade-offs between FSL and fine-tuning, offering insights into how organizations can balance cost and effectiveness for flaky test detection and classification. I’m deeply grateful to my advisor, Dr. Jeremy Bradbury, for his contribution and guidance throughout. The conference will be held in the beautiful city of Napoli, Italy, and I look forward to engaging with fellow researchers, exchanging ideas, and exploring the future of software testing. 🌍 #SoftwareTesting #AI #LLM #FewShotLearning #Research #MachineLearning #ICST2025 #Conference #Italy #Flaky

This paper proposes a channel coding scheme with channel state information of flash memories (i.e., side information of #inter-#cell #interference (#ICI)) to cope with the ICI problem. This side information is obtained before writing data into flash memories and incorporated during the encoding stage. The authors show that flash memories under ICI problem can be transformed into the model of memory with defective cells due to the unique asymmetry property between write (#page #write) and erase (#block #erase) operations. ----Yongjune Kim, Euiseok Hwang, @B. V. K. Vijaya Kumar More details can be found at this link: https://lnkd.in/gJ9awzTM

ASE is the leading research forum in automated software engineering, and MATISSE has organized the 6th ASYDE workshop there! @MATISSE also presented the latest paper, “Towards Synthetic Trace Generation of Modeling Operations using In-Context Learning Approach"

According to Luciano Baresi, a Professor of Software Engineering at Politecnico di Milano, the recently-launched Proceedings of the ACM on Software Engineering (PACMSE) offers a new means to publish novel and top-level research timely in a high-quality journal. In this week's People of ACM profile, Baresi, the EiC of the journal, discusses his goals for PACMSE. He also delves into his research in distributed systems, service-based applications, and the different aspects of mobile, self-adaptive, and pervasive software systems. Read the full interview here: https://bit.ly/3Ulp0Up

Thoughts on applicability (In Software Engineering research) Titus Winters "The gulf between industry and academic interests in Software Engineering appears to continue to widen. With software taking on an ever-growing role in society, it is somewhat concerning that the interests and research focuses of academic software engineers seem to rarely overlap with industry interests.  In this editorial, I hypothesize four major concerns I have with the applicability of much of the research in this area." 1) Solutions without clear or practical context 2) Answering research questions that nobody is asking 3) Small audiences or rare applicability 4) Academic toys vs industry projects Read here: https://lnkd.in/eCSESt4Q

Question to my software engineering research colleagues: what are, for you, the most interesting recent research articles (say, last 5 years) related to software package/library dependency networks/ecosystems and why? I have such a list myself, but prefer not to share it in oder not to bias your responses. (My goal is to update my software evolution lecture for cs masters on this topic, that i am happy to share with anyone interested.)

👨💻 For the software engineering researchers in my network! 🔍 Inferring causal relationships between variables from observational data is a common goal in empirical software engineering ⚡ However, endogeneity due to confounding variables often makes such inferences difficult 💡 In our new article in TOSEM, Stefan Wagner and I introduce a method to clean up such confounding to the field of software engineering: instrumental variable regression 📜 Paper in open access here: https://lnkd.in/e-yiHzry 🙏 Thanks to Bogdan Vasilescu for being an exceptionally helpful Associate Editor on this paper!

Leetcode Daily Problem!! Given two integers n and k, return the kth lexicographically smallest integer in the range [1, n]. Problem Link:https://lnkd.in/g2kk5PpN

✨Here's a little-known fact: In the early days of software testing, engineers would use actual bugs found in hardware components, like a moth, to symbolize defects in the software. This practice led to the coining of the term 'debugging' when removing errors from code. It's a quirky piece of history that showcases the inventive problem-solving spirit of the tech industry! 🐞 The term "debugging" indeed originated from an incident involving an actual moth found in a relay of the Harvard Mark II computer in 1947. This event was documented by computer scientist Grace Hopper, who taped the moth into the computer's logbook with the notation "First actual case of bug being found." This incident led to the term "debugging" being used to describe the process of fixing errors in computer systems. While the specific term "debugging" may not have been widely known, the story behind it is a fascinating piece of computing history. #SoftwareTesting #TechTrivia #Innovation #Debugging