Towards an Automatic Optimisation Model Generator Assisted with Generative Pre-trained Transformer

[Boris Almonacid]

Hello everyone, I've done some recent research presenting an approach to automatically build optimisation models by using a large language model (GPT) 🙂.

⚙ In this process, the user provides input instructions specifying the desired features of the optimisation model. These instructions are then used by the #GPT Agent to generate the #optimisation model. Once the model is created, it is passed on to the optimisation Agent for compilation and execution. If the optimisation agent successfully compiles and solves the model, both the optimised model and its solution are provided to the user as an output.

🧙‍♂️ However, an #automatic #fixing #process is triggered when the optimization model contains compilation errors. This process uses the error message to provide feedback and resolves the problems within the model. By doing so, any inconsistencies or errors in the generated models are promptly addressed, ensuring their accuracy and effectiveness.

📈 To evaluate the effectiveness of the approach, I conducted tests in which the system generated models in the #Minizinc Language for different problem scenarios, evaluating their performance against the desired specifications. The results were promising, showing that my system allows the generation of models that meet the required criteria.

📋 However, I should note that some models may still require further refinement, although it has enormous potential to simplify the modelling process and facilitate troubleshooting across many industries.

The article can be found at [2305.05811] Towards an Automatic Optimisation Model Generator Assisted with Generative Pre-trained Transformer (arxiv.org)

Regards

Boris L. Almonacid

[Gupta, Gopal]

*** Call for Nomination ***

TL;DR: If you or your group have a recent practical accomplishment that highlights the benefits of Prolog-inspired computing, then please consider nominating yourself/your-group for the 2023 Alain Colmerauer Prize. Submission is 3 pages max; self-nomination is required.

The 2023 Alain Colmerauer Prize 

------------------------------- 

Organized by the Association for Logic Programming 

In the summer of 1972, Alain Colmerauer and his team in Marseille developed and implemented the first version of the logic programming language Prolog. Together with both earlier and later collaborations with Robert Kowalski and his colleagues in Edinburgh, this work laid the practical and theoretical foundations for the Prolog language and logic programming of today. Prolog and its related technologies soon became key tools of symbolic programming and Artificial Intelligence. 

2022 was celebrated as the Year of Prolog to recognize the 50th anniversary of these events and highlight the continuing significance of Prolog and Logic Programming both for symbolic, explainable AI, and for computing more generally. The celebration culminated with the award of the inaugural edition of the ALP Alain Colmerauer Prolog Heritage Prize (in short: the Alain Colmerauer Prize) for recent practical accomplishments that highlight the benefits of Prolog-inspired computing for the future. The Alain Colmerauer Prize and the Year of Prolog were organized by the Association for Logic Programming (ALP) and the Prolog Heritage Association (PHA). Details of the 2022 Alain Colmerauer Prize can be found at https://prologyear.logicprogramming.org/ColmerauerPrize.html.

The 2023 Award 

-------------- 

The Alain Colmerauer Prize will continue to be awarded in 2023 and beyond. Nominations are sought for the 2023 edition of the Alain Colmerauer Prize.

Eligibility 

----------- 

Any individual or group of individuals can nominate themselves or their institution(s)/organization(s) for the Prize. Nominations should describe work that meets the purpose of the Prize. Submissions that address the well-being of society or of the planet are especially welcome. 

Submissions 

----------- 

Nominations should explain the contribution and argue for its present and future significance. The submissions must not exceed three pages plus references and may optionally be accompanied by up to two letters of support no longer than 500 words each. Submissions should be made by the candidates themselves, in pdf, through EasyChair at https://easychair.org/conferences/?conf=acprize2023.

Selection and award 

------------------- 

The Prize is given for depth, novelty, and proven or potential impact. The winner is selected by the Jury from the submitted nominations in consultation with the Year of Prolog Scientific Committee. Furthermore, a shortlist of up to five finalists may also be selected in the process. The Jury will provide a detailed citation that explains the basis of the awarding of the Prize. 

The winner receives a certificate and cash support of up to 2,000 Euros for attending the conference and award ceremony. If there are multiple winners, this amount is shared. Finalists also receive certificates.

Timeline for the 2023 award 

--------------------------- 

*******Deadline for submissions******* 

       May 31, 2023.

*******Notification of the shortlisted candidates******* 

       June 10, 2023.

Award and presentation of the 2023 Prize 

----------------------------------------- 

The winner of the 2023 Prize will be announced at the 39th International Conference on Logic Programming (ICLP 2023). 

The 2023 AC Prize Jury 

------------- 

Maria Garcia de la Banda, Gopal Gupta (chair), Manuel Hermenegildo, Y. Annie Liu, and Marie-Christine Rousset 

For more information and details, click here.