CAIN 2022
Mon 16 - Tue 17 May 2022
co-located with ICSE 2022

Welcome to CAIN’22 - 1st Conference on AI Engineering – Software Engineering for AI

May 16-17, 2022 on-line and in-person May 24, 2022 in Pittsburgh, PA, USA

The aim of CAIN’22 is to bring together researchers and practitioners in software engineering, in data-science and AI, and to build up a community that will target the new challenges emerging in Software Engineering for AI-enabled systems.

In development and implementation of AI-enabled systems, the main challenge is not to develop the best models/algorithms, but to provide support for the entire lifecycle – from a business idea, through collection, training, and management of data, code development, product deployment and operation, and its maintenance and evolution. There is a clear need for specific support of Software Engineering for AI.

CAIN has the following goals for the next few years.

  • Identify the main challenges of AI engineering, from an AI and SE perspective, considering industrial needs and their current experiences;
  • Create a roadmap capturing the research directions in AI engineering in relation to AI-based systems lifecycle;
  • Contribute to a better understating of practical problems and understanding differences in approaches of data science and AI/ML and SE;
  • Identify industrial challenges in building and using AI-enabled systems and contribute to solving them
  • Build a thriving community of SE, and data-science and AI practitioners and researchers.

CAIN’22 will have all sessions in a single track. The following sessions are planned:

  • Invited keynotes and panels
  • Presentation of accepted papers
  • Industrial Talks
  • Poster presentations
Dates
You're viewing the program in a time zone which is different from your device's time zone change time zone

Mon 16 May

Displayed time zone: Eastern Time (US & Canada) change

06:00 - 06:30
Welcome to CAINCAIN 2022 at CAIN main room
Chair(s): Jan Bosch Chalmers University of Technology, Jane Cleland-Huang University of Notre Dame, Helena Holmström Olsson Malmö University, Iva Krasteva Sofia University, GATE Institute, Lucy Ellen Lwakatare University of Helsinki, Henry Muccini University of L'Aquila, Italy, Ipek Ozkaya Carnegie Mellon Software Engineering Institute, Thomas Zimmermann Microsoft Research
06:30 - 07:30
Quality AssuranceCAIN 2022 at CAIN main room
Chair(s): Henry Muccini University of L'Aquila, Italy
06:30
15m
Research paper
What is Software Quality for AI Engineers? Towards a Thinning of the FogResearch Paper
CAIN 2022
Valentina Golendukhina University of Innsbruck, Valentina Lenarduzzi University of Oulu, Michael Felderer University of Innsbruck
06:45
15m
Research paper
Exploring ML testing in practice - Lessons learned from an interactive rapid review with Axis CommunicationsResearch Paper
CAIN 2022
Qunying Song Lund University, Markus Borg RISE Research Institutes of Sweden, Emelie Engstrom Lund University, Håkan Ardö Axis Communications, Lund, Sweden, Sergio Rico Lund University, Sweden
Pre-print
07:00
15m
Research paper
Quality Assurance of Generative Dialog Models in an Evolving Conversational Agent Used for Swedish Language PracticeResearch Paper
CAIN 2022
Markus Borg RISE Research Institutes of Sweden, Johan Bengtsson Lund University, Harald Österling Lund University, Alexander Hagelborn NordAxon AB, Isabella Gagner NordAxon AB, Piotr Tomaszewski RISE Research Institutes of Sweden
07:15
15m
Other
Discussion on Quality Assurance
CAIN 2022

07:30 - 07:45
07:45 - 09:15
PostersCAIN 2022 at CAIN main room
Chair(s): Helena Holmström Olsson Malmö University, Iva Krasteva Sofia University, GATE Institute
07:45
30m
Other
Activity: Networking Shuffle
CAIN 2022

08:15
3m
Poster
MLOps: Five Steps to Guide its Effective ImplementationPoster
CAIN 2022
08:18
3m
Poster
Towards A Methodological Framework for Production-ready AI-based Software ComponentsPoster
CAIN 2022
Markus Haug University of Stuttgart, Institute of Software Engineering, Empirical Software Engineering Group, Justus Bogner University of Stuttgart, Institute of Software Engineering, Empirical Software Engineering Group
08:21
3m
Poster
Preliminary Insights to enable automation of the Software Development Process in Software StartUps. A Investigation Study from the use of Artificial Intelligence and Machine LearningPoster
CAIN 2022
Olimar Borges PUCRS University, Valentina Lenarduzzi University of Oulu, Rafael Prikladnicki School of Technology at PUCRS University
08:24
3m
Poster
Identification of Out-of-Distribution Cases of CNN using Class-Based Surprise AdequacyPoster
CAIN 2022
Mira Marhaba Polytechnique Montreal, Ettore Merlo Polytechnique Montreal, Foutse Khomh Polytechnique Montréal, Giuliano Antoniol Polytechnique Montréal
08:27
3m
Poster
Robust Active Learning: Sample-Efficient Training of Robust Deep Learning ModelsPoster
CAIN 2022
Yuejun GUo Interdisciplinary Centre for Security, Qiang Hu University of Luxembourg, Maxime Cordy University of Luxembourg, Luxembourg, Mike Papadakis University of Luxembourg, Luxembourg, Yves Le Traon University of Luxembourg, Luxembourg
08:30
3m
Poster
A New Approach for Machine Learning Security Risk AssessmentPoster
CAIN 2022
Jun Yajima Fujitsu Limited, Maki Inui Fujitsu Limited, Takanori Oikawa Fujitsu Limited, Fumiyoshi Kasahara Fujitsu Limited, Ikuya Morikawa Fujitsu Limited, Nobukazu Yoshioka Waseda University, Japan
File Attached
08:33
3m
Poster
TopSelect: A Topology-based Feature Selection Method for Industrial Machine LearningPoster
CAIN 2022
Hadil Abukwaik ABB Corporate Research, Lefter Sula ABB Corporate Research Center, Pablo Rodriguez ABB
08:36
3m
Poster
Pynblint: a Static Analyzer for Python Jupyter NotebooksPoster
CAIN 2022
Luigi Quaranta University of Bari, Italy, Fabio Calefato University of Bari, Filippo Lanubile University of Bari
Pre-print File Attached
08:39
3m
Poster
Traceable Business-to-Safety Analysis Framework for Safety-critical Machine Learning SystemsPoster
CAIN 2022
Jati Hiliamsyah Husen Waseda University, Hironori Washizaki Waseda University, Hnin Thandar Tun Waseda University, Nobukazu Yoshioka Waseda University, Japan, Hironori Takeuchi Musashi University, Yoshiaki Fukazawa Waseda University
Media Attached File Attached
08:42
3m
Poster
Structural Causal Models as Boundary Objects in AI System DevelopmentPoster
CAIN 2022
Hans-Martin Heyn University of Gothenburg & Chalmers University of Technology, Eric Knauss Chalmers | University of Gothenburg
08:45
30m
Other
Poster Visits
CAIN 2022

09:15 - 09:30
09:30 - 11:00
Training & LearningCAIN 2022 at CAIN main room
Chair(s): Jan Bosch Chalmers University of Technology
09:30
15m
Research paper
An Empirical Evaluation of Flow Based Programming in the Machine Learning Deployment ContextResearch Paper
CAIN 2022
Andrei Paleyes Department of Computer Science and Technology, Univesity of Cambridge, Christian Cabrera Department of Computer Science and Technology, Univesity of Cambridge, Neil D. Lawrence Department of Computer Science and Technology, Univesity of Cambridge
Pre-print Media Attached
09:45
15m
Research paper
Automatic Checkpointing and Deterministic Training for Deep LearningResearch Paper
CAIN 2022
Xiangzhe Xu Purdue University, Hongyu Liu Huawei Galois Lab, China, Guanhong Tao Purdue University, USA, Zhou Xuan Purdue University, Xiangyu Zhang Purdue University
10:00
15m
Research paper
Influence-Driven Data Poisoning in Graph-Based Semi-Supervised ClassifiersResearch Paper
CAIN 2022
Adriano Franci University of Luxembourg, Maxime Cordy University of Luxembourg, Luxembourg, Martin Gubri University of Luxembourg, Luxembourg, Mike Papadakis University of Luxembourg, Luxembourg, Yves Le Traon University of Luxembourg, Luxembourg
10:15
15m
Industry talk
Engineering a Platform for Reinforcement Learning WorkloadsIndustry Talk
CAIN 2022
Ali Kanso Microsoft, Kinshuman Patra Microsoft
10:30
15m
Research paper
Method Cards for Prescriptive Machine-Learning TransparencyResearch Paper
CAIN 2022
David Adkins Meta AI, Bilal Alsallakh Meta AI, Adeel Cheema Meta AI, Narine Kokhlikyan Meta AI, Emily McReynolds Meta AI, Pushkar Mishra Meta AI, Chavez Procope Meta AI, Jeremy Sawruk Meta AI, Erin Wang Meta AI, Polina Zvyagina Meta AI
10:45
15m
Other
Discussion on Training & Learning
CAIN 2022

11:00 - 11:15

Tue 17 May

Displayed time zone: Eastern Time (US & Canada) change

06:00 - 07:30
AI Engineering PracticesCAIN 2022 at CAIN main room
Chair(s): Ipek Ozkaya Carnegie Mellon Software Engineering Institute
06:00
15m
Research paper
Towards a Roadmap for Software Engineering for Responsible AIResearch Paper
CAIN 2022
Qinghua Lu CSIRO’s Data61, Liming Zhu CSIRO’s Data61; UNSW, Xiwei (Sherry) Xu CSIRO Data61, Jon Whittle CSIRO's Data61 and Monash University, Zhenchang Xing Australian National University
06:15
15m
Research paper
AI Governance in the System Development Life Cycle: Insights on Responsible Machine Learning EngineeringResearch Paper
CAIN 2022
Samuli Laato University of Turku, Teemu Birkstedt University of Turku, Matti Mäntymäki University of Turku, Matti Minkkinen University of Turku, Tommi Mikkonen University of Helsinki
06:30
15m
Research paper
The Goldilocks Framework: Towards Selecting the Optimal Approach to Conducting AI ProjectsResearch Paper
CAIN 2022
Rimma Dzhusupova McDermott, Jan Bosch Chalmers University of Technology, Helena Holmström Olsson Malmö University
06:45
15m
Research paper
What Is an AI Engineer? An Empirical Analysis of Job Ads in the NetherlandsResearch Paper
CAIN 2022
Marcel Meesters Fontys University of Applied Sciences, Petra Heck Fontys University of Applied Sciences, Alexander Serebrenik Eindhoven University of Technology
Pre-print
07:00
15m
Research paper
Data is about Detail: An Empirical Investigation for Software Systems with NLP at CoreResearch Paper
CAIN 2022
Anmol Singhal TCS Research, Preethu Rose Anish TCS Research, Pratik Sonar TCS Research, Smita Ghaisas TCS Research
File Attached
07:15
15m
Other
Discussion on AI Engineering Practices
CAIN 2022

07:30 - 07:45
07:45 - 09:15
AI Models & PipelinesCAIN 2022 at CAIN main room
Chair(s): Lucy Ellen Lwakatare University of Helsinki
07:45
15m
Industry talk
Practical Insights of Repairing Model Problems on Image ClassificationIndustry Talk
CAIN 2022
Akihito Yoshii Fujitsu Limited, Susumu Tokumoto Fujitsu Limited, Fuyuki Ishikawa National Institute of Informatics
08:00
15m
Research paper
UDAVA: An Unsupervised Learning Pipeline for Sensor Data Validation in ManufacturingResearch Paper
CAIN 2022
Erik Johannes Husom SINTEF Digital, Simeon Tverdal SINTEF Digital, Arda Goknil SINTEF Digital, Sagar Sen
08:15
15m
Research paper
Black-Box Models for Non-Functional Properties of AI Software SystemsResearch Paper
CAIN 2022
Daniel Friesel Universität Osnabrück, Olaf Spinczyk Universität Osnabrück
DOI Pre-print
08:30
15m
Research paper
Improving Generalizability of ML-enabled Software through Domain SpecificationResearch Paper
CAIN 2022
Hamed Barzamini , Mona Rahimi Northern Illinois University, Murtuza Shahzad Northern Illinois University, Hamed Alhoori Northern Illinois University
08:45
15m
Research paper
Data Sovereignty for AI Pipelines: Lessons Learned from an Industrial Project at Mondragon CorporationResearch Paper
CAIN 2022
Marcel Altendeitering Fraunhofer ISST, Julia Pampus Fraunhofer ISST, Felix Larrinaga Mondragon Unibertsitatea, Jon Legaristi Mondragon Unibertsitatea, Falk Howar TU Dortmund University
File Attached
09:00
15m
Other
Discussion on AI Models & Pipelines
CAIN 2022

09:15 - 09:30
09:30 - 11:00
AI SmellsCAIN 2022 at CAIN main room
Chair(s): Ipek Ozkaya Carnegie Mellon Software Engineering Institute, Thomas Zimmermann Microsoft Research
09:30
30m
Other
Activity: Brainwriting
CAIN 2022

10:00
15m
Research paper
Code Smells for Machine Learning ApplicationsResearch Paper
CAIN 2022
Haiyin Zhang AI for Fintech Research, ING, Luís Cruz Deflt University of Technology, Arie van Deursen Delft University of Technology, Netherlands
Pre-print
10:15
15m
Research paper
Data Smells: Categories, Causes and Consequences, and Detection of Suspicious Data in AI-based SystemsResearch Paper
CAIN 2022
Harald Foidl University of Innsbruck, Michael Felderer University of Innsbruck, Rudolf Ramler Software Competence Center Hagenberg
Pre-print
10:30
15m
Research paper
Data smells in Public DatasetsResearch Paper
CAIN 2022
Arumoy Shome Delft University of Technology, Luís Cruz Deflt University of Technology, Arie van Deursen Delft University of Technology, Netherlands
Pre-print
10:45
15m
Other
Discussion on Smells in AI
CAIN 2022

11:00 - 11:15
11:15 - 12:15
Keynote Saleema AmershiCAIN 2022 at CAIN main room
Chair(s): Jane Cleland-Huang University of Notre Dame
11:15
60m
Keynote
Challenges in creating responsible and human-centered AI
CAIN 2022
Saleema Amershi Microsoft Research
12:15 - 12:30
Closing and AwardsCAIN 2022 at CAIN main room
Chair(s): Jan Bosch Chalmers University of Technology, Jane Cleland-Huang University of Notre Dame, Helena Holmström Olsson Malmö University, Iva Krasteva Sofia University, GATE Institute, Lucy Ellen Lwakatare University of Helsinki, Henry Muccini University of L'Aquila, Italy, Ipek Ozkaya Carnegie Mellon Software Engineering Institute, Thomas Zimmermann Microsoft Research

Accepted Papers

Title
AI Governance in the System Development Life Cycle: Insights on Responsible Machine Learning EngineeringResearch Paper
CAIN 2022
An Empirical Evaluation of Flow Based Programming in the Machine Learning Deployment ContextResearch Paper
CAIN 2022
Pre-print Media Attached
A New Approach for Machine Learning Security Risk AssessmentPoster
CAIN 2022
File Attached
Automatic Checkpointing and Deterministic Training for Deep LearningResearch Paper
CAIN 2022
Black-Box Models for Non-Functional Properties of AI Software SystemsResearch Paper
CAIN 2022
DOI Pre-print
Code Smells for Machine Learning ApplicationsResearch Paper
CAIN 2022
Pre-print
Data is about Detail: An Empirical Investigation for Software Systems with NLP at CoreResearch Paper
CAIN 2022
File Attached
Data Smells: Categories, Causes and Consequences, and Detection of Suspicious Data in AI-based SystemsResearch Paper
CAIN 2022
Pre-print
Data smells in Public DatasetsResearch Paper
CAIN 2022
Pre-print
Data Sovereignty for AI Pipelines: Lessons Learned from an Industrial Project at Mondragon CorporationResearch Paper
CAIN 2022
File Attached
Engineering a Platform for Reinforcement Learning WorkloadsIndustry Talk
CAIN 2022
Exploring ML testing in practice - Lessons learned from an interactive rapid review with Axis CommunicationsResearch Paper
CAIN 2022
Pre-print
Identification of Out-of-Distribution Cases of CNN using Class-Based Surprise AdequacyPoster
CAIN 2022
Improving Generalizability of ML-enabled Software through Domain SpecificationResearch Paper
CAIN 2022
Influence-Driven Data Poisoning in Graph-Based Semi-Supervised ClassifiersResearch Paper
CAIN 2022
Method Cards for Prescriptive Machine-Learning TransparencyResearch Paper
CAIN 2022
MLOps: Five Steps to Guide its Effective ImplementationPoster
CAIN 2022
Practical Insights of Repairing Model Problems on Image ClassificationIndustry Talk
CAIN 2022
Preliminary Insights to enable automation of the Software Development Process in Software StartUps. A Investigation Study from the use of Artificial Intelligence and Machine LearningPoster
CAIN 2022
Pynblint: a Static Analyzer for Python Jupyter NotebooksPoster
CAIN 2022
Pre-print File Attached
Quality Assurance of Generative Dialog Models in an Evolving Conversational Agent Used for Swedish Language PracticeResearch Paper
CAIN 2022
Robust Active Learning: Sample-Efficient Training of Robust Deep Learning ModelsPoster
CAIN 2022
Structural Causal Models as Boundary Objects in AI System DevelopmentPoster
CAIN 2022
The Goldilocks Framework: Towards Selecting the Optimal Approach to Conducting AI ProjectsResearch Paper
CAIN 2022
TopSelect: A Topology-based Feature Selection Method for Industrial Machine LearningPoster
CAIN 2022
Towards A Methodological Framework for Production-ready AI-based Software ComponentsPoster
CAIN 2022
Towards a Roadmap for Software Engineering for Responsible AIResearch Paper
CAIN 2022
Traceable Business-to-Safety Analysis Framework for Safety-critical Machine Learning SystemsPoster
CAIN 2022
Media Attached File Attached
UDAVA: An Unsupervised Learning Pipeline for Sensor Data Validation in ManufacturingResearch Paper
CAIN 2022
What Is an AI Engineer? An Empirical Analysis of Job Ads in the NetherlandsResearch Paper
CAIN 2022
Pre-print
What is Software Quality for AI Engineers? Towards a Thinning of the FogResearch Paper
CAIN 2022

1. Call for Papers

Keynotes

  • Chris Re: Software 2.0, Foundation Models, Data-Centric AI, and why I’m excited enough to tolerate these buzzwords.
  • Saleema Amershi: Challenges in Creating Responsible and Human-cCentered AI

Accepted contributions

For the list of accepted papers, posters, and industry talks, please visit this link

Welcome to CAIN’22 - 1st International Conference on AI Engineering – Software Engineering for AI

Call for submission - research and experience papers

You are invited to submit research and experience papers describing the challenges, new research results, visionary ideas, or experience papers from or in cooperation with practitioners.

Topics of interests

The overall area is Software Engineering for AI, i.e. means to improve development of software AI-based systems and software-intensive systems, including topics relevant for the entire lifecycle. The suggested topics are (but not limited to):

  • System and software requirements and their relations AI/ML modelling;
  • Data management ensuring relevance and efficiency related to business goals;
  • System and software architecture of AI-based systems;
  • Integration of AI-development process and software development processes, including continuous and federated ML, continuous deployment, system and software evolution;
  • Ensuring and managing system and software nonfunctional properties and their relation to AI/ML properties, including run-time properties such as performance, safety, security, reliability, and life-cycle properties including reusability, maintainability and evolution;
  • Development teams, organizational and management issues for a successful development of AI-systems.

Review criteria

The paper submissions will undergo a review process with three independent reviews and a virtual PC decision meeting. The acceptance criteria include novelty, research and industrial relevance, soundness, experiences, and preliminary results.

Submission form

The papers should be submitted to HotCRP web page (https://cain2022.hotcrp.com/). The submission deadline is firm.

Update (1/17/2022) Submissions must be registered by January 20 with the paper title, author list and some additional metadata. The deadline for the completed paper is January 25.

All submissions must adhere to the following requirements:

  • The page limit is 10 pages plus 2 additional pages of references.
  • Submissions must be unpublished original work and should not be under review or submitted elsewhere while being under consideration.
  • The submissions must adhere to the rules specified by the Submission format.
  • The submissions will be reviewed in a form of Double Blind Review process, i.e. the authors’ names must be omitted from the submission. For formatting and additional instructions see Submission format.

Publication

The accepted papers will be published in the ICSE 2022 Co-located Event Proceedings and included in the IEEE and ACM Digital Libraries. Authors of accepted papers are required to register and present their accepted paper at the conference in order for the paper to be included in the proceedings and the Digital Libraries.

The official publication date is the date the proceedings are made available in the ACM or IEEE Digital Libraries. This date may be up to two weeks prior to the first day of ICSE 2022. The official publication date affects the deadline for any patent filings related to published work.

Submission format

All submissions to CAIN 2022 must adhere to the following rules:

  • Submissions must strictly conform to the ACM formatting instructions for both LaTeX and Word users. LaTeX users must use the provided acmart.cls and ACM-Reference-Format.bst without modification, enable the conference format in the preamble of the document (i.e., {\documentclass[sigconf,review]{acmart}), and use the ACM reference format for the bibliography (i.e., \bibliographystyle{ACM-Reference-Format}). The review option adds line numbers, thereby allowing referees to refer to specific lines in their comments.
  • All submissions must be in PDF.
  • By submitting to CAIN, authors acknowledge that they are aware of and agree to be bound by the ACM Policy and Procedures on Plagiarism and the IEEE Plagiarism FAQ. The authors also acknowledge that they conform to the authorship policy of the ACM and the authorship policy of the IEEE.
  • The CAIN 2020 Full paper review will employ a double-anonymous review process. Thus, no submission may reveal its authors’ identities. The authors must make every effort to honor the double-anonymous review process. In particular:
    • Authors’ names must be omitted from the submitted paper. LaTeX users can add anonymous (omitting author names) option (i.e. \documentclass[sigconf,review,anonymous]{acmart} specification).
    • All references to the author’s prior work should be in the third person.
    • Authors are encouraged to title their submission differently than preprints of the authors on ArXiV or similar sites. During review, authors should not publicly use the submission title.
  • The review process of posters and Industrial Talks proposals will use a single blind review process, i.e. the authors should specify their names in the submitted document.

3. Call for posters

In addition to full technical papers, CAIN 2022 provides the opportunity to submit posters which will be included in the proceedings as two-page extended abstracts. New ideas, a starting work and the results, or presentation of challenges in theory in practice, related to the topics of AI engineering, are welcome to be submitted. In addition, the full papers that will not be accepted, but are of interest for the AI engineering community, will be invited to submit to this track.

The authors need to submit a 2-page extended abstract that should adhere to the ICSE 2022 Conference Proceedings Formatting Guidelines. The 2-page extended abstract of each accepted poster may, at the authors’ discretion, be published in the CAIN’22 proceedings.

The submissions must adhere to the rules specified by the Submission format.

Abstracts must be submitted electronically at the submission site EasyChair CAIN2022-posters. by the submission deadline. A submission will be desk rejected if it does not comply with the instructions and size limits. At least one author of each accepted extended abstract is required to register for the CAIN 2022 conference and to present the poster. Each accepted poster will be presented by its authors during the 2 days of CAIN conference.

2. Call for Industrial Talks

The goal of industrial talks is to share experiences in industrial applications of software engineering of AI-enabled systems and lessons learned from applications of various techniques and practices. This type of submission is only open to individuals willing to share practical lessons learned directly from the field.

We solicit practitioner-oriented talks on topics that are likely to be relevant to both industrial and academic attendees. Talk proposals should include a short abstract (150 words), and up to 8 keywords. In addition, the proposal should include a “talk description”, which describes what the talk will be about, highlighting its key points and the reason why it is relevant and important to the software engineering community (500 words). This description will be included in the CAIN’22 proceedings as a two-page long extended abstract for each industrial talk.

In addition, please include up to 10 slides that represent the content of your talk. Submissions should also include the title, the name and affiliation of each presenter, and a speaker biography. Talk proposals can include supporting supplemental materials such as white papers or videos. Please indicate a desired length of either 15 min or 30 min for your talk.

The papers should be submitted to Easy Chair page. The submission deadline is firm.

The submissions must adhere to the rules specified by the Submission format.

The submitted proposals will be reviewed by the Industrial Track Committee.

Questions? Use the CAIN contact form.