December 2024 - Stack@CS Reading Group

[Sysreading] Research Showcase and Last Stack@CS meeting of this semester
by Kabir, Md Mahir Asef 16 Dec '24

16 Dec '24

Hi all, We will have our last Stack@CS meeting of this semester on 12/18 (Wednesday) at Gilbert Place 4001. In this meeting, we will have a retrospective session and a research showcase. Stack@CS Research Showcase The Stack@CS Center for Systems Research is hosting a Fall 2024 research showcase to highlight all of the research and accomplishments from students this year. The showcase will be Wednesday, 12/18, at 10am in Gilbert 4001. Students are invited to present their work as a lightning talk (5 min presentation) or poster, and contributions should describe innovative and relevant research. Please complete the form<https://docs.google.com/forms/d/e/1FAIpQLSetIvTuyFQ4gYn5BcPTlwXEcqmRNMk1OS8…> if you are interested in presenting for this event and contact Dr. @Brown, Chris (dcbrown(a)vt.edu<mailto:dcbrown@vt.edu>) if you have additional questions. Best, Mahir Md Mahir Asef Kabir PhD Candidate Department of Computer Science Virginia Tech Get Outlook for Android<https://aka.ms/AAb9ysg>

1 0

[Sysreading] Stack@CS Group Meeting this Week & Announcement about Stack@CS Research Showcase
by Kabir, Md Mahir Asef 09 Dec '24

09 Dec '24

Hi all, We will have our next Stack@CS<https://sites.google.com/vt.edu/stackcs-fall24/home> reading group meeting this Wednesday (12/11). In this meeting, Wenxin Jiang (Ph.D. Candidate, Purdue University) will be presenting his work titled "Trustworthy Reuse in Open-Source AI Model Ecosystems: How Far are We?". Location: 220 Gilbert Street (4001) Zoom: https://virginiatech.zoom.us/my/dcbrown Time: 10 am - 11 am Details of the talk and a short bio of our guest speaker are given below - Title: Trustworthy Reuse in Open-Source AI Model Ecosystems: How Far are We? Bio: Wenxin Jiang is a Ph.D. candidate in the Elmore Family School of Electrical & Computer Engineering at Purdue University, under the supervision of Dr. James C. Davis. His research primarily focuses on Software Engineering for AI, with an emphasis on open-source pre-trained AI models. His work adapts methods from studies on traditional software package registries (e.g., NPM, PyPI) to open-source pre-trained AI model (PTM) supply chain. He is also interested in machine learning systems, software supply chain security, and trustworthy/responsible AI. His work has been published at ICSE, FSE, ASE, ISSTA, EMSE, ESEM, MSR, and SCORED. Abstract: Deep neural networks are increasingly integrated into software systems, yet developing these networks from scratch has become challenging as state-of-the-art architectures grow in complexity. Like traditional software engineers, deep learning practitioners are now reusing pre-trained models (PTMs) for downstream tasks and environments. However, unlike traditional software, where reuse practices and challenges have been well-studied, the knowledge foundation for PTM ecosystems remains underdeveloped. My findings indicate that reusing pre-trained models is often complex and carries potential risks, as issues within open-source AI models can propagate downstream, potentially impacting dependent applications. I will conclude by discussing ongoing work aimed at strengthening various aspects of the PTM supply chain, with a focus on enhancing trustworthiness, reusability, and security. Looking forward to seeing you all tomorrow. Hopefully it will be an interesting and interactive session. Please fill out this form<https://forms.office.com/r/FUZsCXvLyK> if you want to be added to the Microsoft Teams team. Stack@CS Research Showcase The Stack@CS Center for Systems Research is hosting a Fall 2024 research showcase to highlight all of the research and accomplishments from students this year. The showcase will be Wednesday, 12/18, at 10am in Gilbert 4001. Students are invited to present their work as a lightning talk (5 min presentation) or poster, and contributions should describe innovative and relevant research. Please complete the form<https://docs.google.com/forms/d/e/1FAIpQLSetIvTuyFQ4gYn5BcPTlwXEcqmRNMk1OS8…> if you are interested in presenting for this event and contact Dr. @Brown, Chris<mailto:dcbrown@vt.edu> (dcbrown(a)vt.edu) if you have additional questions. Best, Mahir Md Mahir Asef Kabir PhD Candidate Department of Computer Science Virginia Tech

1 0

[Sysreading] Stack@CS Reading Group Meeting this Week - Guest Speaker from Microsoft Research
by Kabir, Md Mahir Asef 04 Dec '24

04 Dec '24

Hi all, We will have our next Stack@CS<https://sites.google.com/vt.edu/stackcs-fall24/home> reading group meeting tomorrow, Wednesday (12/04). In this meeting, Dr. Saikat Chakraborty, who is a Senior Researcher at the Research in Software Engineering (RiSE) group at Microsoft Research, will be presenting his work titled "Towards Trusted Automated Programming by Intent Formalization and Proof-Oriented Programming". Location: 220 Gilbert Street (4001) Zoom: https://virginiatech.zoom.us/my/dcbrown Time: 10 am - 11 am Details of the talk and a short bio of our guest speaker are given below - Title: Towards Trusted Automated Programming by Intent Formalization and Proof-Oriented Programming. Abstract: The recent excitement around AI-assisted programming has been tempered by concerns around the trustworthiness of AI-generated code. Languages that offer static guarantees can help reduce some of these concerns, e.g., having AI generate safe Rust code rather than C eliminates the risk of AI-introduced memory safety issues. Taking this line of thinking to its limit, using AI to generate code in proof-oriented languages which allow programs to be augmented with specification and proofs of correctness could eliminate trust in AI-generated code. There are two associated problems in this formulation – (a) formalizing and specifying the intent, and (b) programming in safe proof-oriented language that ensures correctness w.r.t. the intent. In this talk, I will touch bases on how AI can help supercharge in both of these domains, and our vision on proceeding towards Trusted AI programming. Bio: Dr. Saikat Chakraborty is a Senior Researcher at the Research in Software Engineering (RiSE) group at Microsoft Research. His expertise in Programming Language Processing (PLP) -- a coalescence between Software Engineering and Artificial Intelligence. His research is centered around two orthogonal directions -- Reliability in AI-generated program - where AI is the main driver, their research set check and bounds, so that AI does not go astray, and (ii)AI for Reliable Programming - where AI is the assistant helping programmers to write code with associated formal guarantees. His research has been published in different Software Engineering and Artificial Intelligence conferences and journals, including, ICSE, FSE, ASE, ACL, NAACL, EMNLP, TSE, ISSTA, etc. His work received best paper awards from ISSTA’23, SCAM’24. Looking forward to seeing you all tomorrow. Hopefully it will be an interesting and interactive session. Please fill out this form<https://forms.office.com/r/FUZsCXvLyK> if you want to be added to the Microsoft Teams team. Best, Mahir Md Mahir Asef Kabir PhD Candidate Department of Computer Science Virginia Tech

1 1