MAPS 2023 Workshop

Schedule: Dec. 3, 2023

[9:30 AM - 10:30 AM]

Keynote 1: Beyond Code Completion: Towards the Next Generation of AI for SE

Speaker: Vincent Hellendoorn

[10:30 AM -11:00 AM]

Coffee Break

[11:00 AM -12:00 PM]

Session 1

Symmetry-Preserving Program Representations for Learning Code Semantics.

Kexin Pei, Weichen Li, Qirui Jin, Shuyang Liu, Scott Geng, Lorenzo Cavallaro, Junfeng Yang, Suman Jana

InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback

John Yang, Akshara Prabhakar, Karthik Narasimhan, Shunyu Yao

Do Code Generation Models Think Like Us? An Empirical Study on Attention Alignment between Large Language Models and Human Programmers

Bonan Kou, Shengmai Chen, Zhijie Wang, Lei Ma, Tianyi Zhang

Predicting the Success of x86-64 Binary Rewriters

Akshay Sood, Keara Hill, Kimble Houck, Jonathan Dorn, Zak Fry

[1:30 PM -2:30 PM]

Keynote 2: Non-canonical Tasks for Synthesis and Learning

Speaker: Kevin Ellis

[2:30 PM -3:30 PM]

Session 2

Large Language Models Should Ask Clarifying Questions to Increase Confidence in Generated Code

Jie JW Wu

An Empirical Study of Code Generation Errors Made by Large Language Models

Da Song, Zijie Zhou, Zhijie Wang, Yuheng Huang, Shengmai Chen, Bonan Kou, Lei Ma, Tianyi Zhang

Efficient Composition of Data Management and Machine Learning Systems via Common Intermediate Layers

Supun Madusha Bandara Abeysinghe Tennakoon Mudiyanselage, Fei Wang, Gregory Essertel, Tiark Rompf

A Preliminary Evaluation of LLM-Based Fault Localization

Sungmin Kang, Gabin An, Shin Yoo

[3:30 PM -4:00 PM]

Coffee Break

[4:00 PM -5:00 PM]

Discussion

Fishbowl discussion led by Baishakhi Ray and Tianyi Zhang

About MAPS 2023

The integration of programming languages, software engineering, and machine learning has the potential to transform software development by enabling new forms of human-machine collaboration. The recent development of pre-trained large language models for programming has introduced exciting possibilities for automating programming tasks, improving developer productivity, and enhancing code quality. However, there are still significant open research questions that need to be addressed in this emerging field.

The primary motivation of this workshop is to bring together researchers from programming languages, software engineering, and machine learning communities to discuss and explore the latest advances in machine learning models for programming. The workshop aims to provide a platform for researchers to share their ideas, exchange research results, and collaborate on new research directions.

The workshop aims to achieve multiple objectives by bringing together researchers from programming languages, software engineering, and machine-learning communities. Firstly, it aims to foster collaboration between the three communities, encouraging the development of new approaches for software development that can benefit from machine learning models. Secondly, the workshop provides a forum for researchers to present and discuss their latest research findings and ideas, thus enabling the exchange of knowledge and the identification of open research questions. Thirdly, the workshop aims to explore the challenges and opportunities of machine learning models for software development activities, with the goal of identifying new research directions in this emerging field. Fourthly, the workshop encourages the development of new models, algorithms, and tools for programming with machine learning models to improve software development efficiency and effectiveness. Finally, the workshop aims to discuss the ethical and societal implications of machine learning models for programming, promoting responsible development and deployment of this technology.

To achieve these objectives, the workshop will feature a combination of keynote talks, research paper presentations, panel discussions, and open discussions. The workshop will provide ample opportunities for researchers to interact, discuss, and collaborate on new research directions. We believe that this workshop will significantly contribute to the advancement of machine learning models for programming and will foster cross-disciplinary research between programming languages, software engineering, and machine-learning communities.

Travel Support

MAPS 2023 Workshop provides the travel support for paticipants through NSF travel awards. See Travel Support for details