Prompt Evolution for Generative AI: A Classifier-Guided Approach

Abhishek Gupta; Caishun Chen; Kavitesh K. Bali; Melvin Wong; Yew-Soon Ong

arxiv: 2305.16347 · v2 · submitted 2023-05-24 · 💻 cs.LG · cs.AI· cs.CV· cs.NE

Prompt Evolution for Generative AI: A Classifier-Guided Approach

Melvin Wong , Yew-Soon Ong , Abhishek Gupta , Kavitesh K. Bali , Caishun Chen This is my paper

Pith reviewed 2026-05-24 08:57 UTC · model grok-4.3

classification 💻 cs.LG cs.AIcs.CVcs.NE

keywords prompt evolutiongenerative AIevolutionary algorithmsimage generationmulti-objective optimizationclassifier guidancePareto optimization

0 comments

The pith

Evolutionary selection and variation during image generation produces outputs more faithful to user preferences via classifier guidance.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces prompt evolution as a process that applies evolutionary selection pressure and variation inside the generative AI loop itself. It demonstrates a concrete multi-objective version that treats labels from a pre-trained multi-label image classifier as simultaneous optimization targets. The method uses the generative model's own stochastic sampling as an implicit mutation operator to generate candidate images and retain those that best satisfy the combined objectives. This approach aims to close the gap between prompt wording and actual output content without requiring changes to the underlying model or manual prompt rewriting.

Core claim

Prompt evolution imparts evolutionary selection pressure and variation during the generative process to produce multiple outputs that satisfy the target concepts and preferences better; a multi-objective instantiation uses predicted classifier labels as objectives and the pre-trained generative model's stochastic capability as implicit mutation to automate creation of Pareto-optimized images more faithful to user preferences.

What carries the argument

Classifier-guided multi-objective evolutionary algorithm that treats the generative model's stochastic sampling as implicit mutation operators to evolve populations toward Pareto fronts defined by classifier label scores.

If this is right

Multiple candidate images can be produced that trade off different preference dimensions without manual intervention.
The generative model itself supplies the variation needed for evolution, removing the need to design explicit mutation operators.
Diversified outputs can be retained that collectively cover a wider range of user-specified concepts than single-pass generation.
The same selection mechanism can be applied at inference time to any prompt without retraining or fine-tuning the base model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The approach could be extended to text or audio generation by swapping in suitable classifiers or reward models for those domains.
Interactive versions might allow users to supply direct preference feedback instead of relying solely on pre-trained classifiers.
The method suggests a general pattern for combining evolutionary computation with any stochastic generative model that produces variable outputs from the same conditioning.

Load-bearing premise

Predicted labels from the classifiers accurately represent the user preferences implied by the prompts and can be treated as optimizable objectives.

What would settle it

A controlled comparison in which human raters score faithfulness of images produced by the evolutionary process versus standard sampling from the same model, checking whether the evolutionary outputs receive reliably higher preference scores.

Figures

Figures reproduced from arXiv: 2305.16347 by Abhishek Gupta, Caishun Chen, Kavitesh K. Bali, Melvin Wong, Yew-Soon Ong.

**Figure 2.** Figure 2: Comparison of brute force approach versus Prompt [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Figure 3: Prompt derived from the proverb: “The family that plays together stays together”. The nondominated population of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

read the original abstract

Synthesis of digital artifacts conditioned on user prompts has become an important paradigm facilitating an explosion of use cases with generative AI. However, such models often fail to connect the generated outputs and desired target concepts/preferences implied by the prompts. Current research addressing this limitation has largely focused on enhancing the prompts before output generation or improving the model's performance up front. In contrast, this paper conceptualizes prompt evolution, imparting evolutionary selection pressure and variation during the generative process to produce multiple outputs that satisfy the target concepts/preferences better. We propose a multi-objective instantiation of this broader idea that uses a multi-label image classifier-guided approach. The predicted labels from the classifiers serve as multiple objectives to optimize, with the aim of producing diversified images that meet user preferences. A novelty of our evolutionary algorithm is that the pre-trained generative model gives us implicit mutation operations, leveraging the model's stochastic generative capability to automate the creation of Pareto-optimized images more faithful to user preferences.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper sketches a conceptual evolutionary approach to refining prompts during generation via classifier guidance and implicit model mutations, but the abstract alone gives no methods or results to judge if it works.

read the letter

The main takeaway is that this paper puts forward a conceptual method for evolving prompts on the fly during image generation, using pre-trained multi-label classifiers to guide a multi-objective evolutionary process, with the generative model's stochastic outputs acting as mutations. What is new is the specific instantiation of evolutionary algorithms in this setting, where selection happens based on classifier predictions to produce Pareto fronts of images that better satisfy user preferences. The paper does a good job explaining how this differs from prior work on prompt enhancement before generation or model fine-tuning. It also correctly notes that the implicit mutation from the model itself is a practical advantage, avoiding the need to design explicit variation operators. The approach is grounded in existing tools like classifiers and generative models, which is a plus for reproducibility if implemented. However, the soft spots are that the abstract gives no algorithm details, no pseudocode, and no experimental results to back up the claims. The key assumption that classifier labels serve as faithful objectives for user preferences is stated but not justified or tested, and this could be a real issue for complex or subjective preferences where classifiers often fail. Without any validation, it's impossible to know if the method actually improves outputs or just adds overhead. The citation pattern isn't an issue here since it's a new proposal. This kind of paper is for researchers exploring evolutionary computation applied to generative models or prompt optimization techniques. Someone already working in those areas might find the framing useful for thinking about new experiments. Overall, I don't think it deserves peer review yet because there's not enough substance to review beyond the high-level idea. If the authors add concrete methods and some results, it could be worth considering.

Referee Report

1 major / 0 minor

Summary. The paper proposes 'prompt evolution' as a new paradigm for generative AI, in which evolutionary selection pressure and variation are applied during the generative process itself (rather than only before or after) to produce outputs that better satisfy target concepts and user preferences implied by prompts. It instantiates this idea via a multi-objective evolutionary algorithm guided by pre-trained multi-label image classifiers whose predicted labels serve as objectives; the pre-trained generative model supplies implicit mutation via its stochastic sampling, with the aim of automatically yielding diversified, Pareto-optimal images.

Significance. If the approach can be shown to work, it would constitute a distinct post-prompt, model-agnostic mechanism for preference alignment that does not require retraining or prompt rewriting. The use of the generative model itself for mutation and the framing of classifier outputs as explicit multi-objective signals are conceptually interesting. However, the manuscript supplies no algorithmic specification, pseudocode, or empirical results, so the practical significance cannot be assessed.

major comments (1)

[Abstract] Abstract: the central claim that the classifier-guided multi-objective evolutionary process yields 'Pareto-optimized images more faithful to user preferences' is unsupported by any derivation, algorithm description, or validation. The manuscript consists solely of the abstract and therefore provides no evidence that the predicted classifier labels can be optimized in this manner or that the implicit mutation produces the claimed improvement.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their review and for noting the conceptual interest in framing prompt evolution as a post-prompt, model-agnostic alignment mechanism. We address the single major comment below.

read point-by-point responses

Referee: [Abstract] Abstract: the central claim that the classifier-guided multi-objective evolutionary process yields 'Pareto-optimized images more faithful to user preferences' is unsupported by any derivation, algorithm description, or validation. The manuscript consists solely of the abstract and therefore provides no evidence that the predicted classifier labels can be optimized in this manner or that the implicit mutation produces the claimed improvement.

Authors: The referee correctly observes that the submitted manuscript consists only of the abstract and therefore supplies neither an algorithmic specification nor empirical results. This version was prepared as a concise conceptual outline to introduce the prompt-evolution paradigm and the use of pre-trained multi-label classifiers as explicit multi-objective signals. We agree that the central claim requires substantiation and will revise the manuscript to include (i) a precise description of the multi-objective evolutionary algorithm, (ii) pseudocode showing how the generative model’s stochastic sampling supplies implicit mutation, and (iii) preliminary experiments that track classifier-label improvement across generations and demonstrate Pareto fronts aligned with user-specified preferences. revision: yes

Circularity Check

0 steps flagged

No significant circularity; method relies on external pre-trained components

full rationale

The paper presents a conceptual framework for prompt evolution via classifier-guided multi-objective optimization during generation. It explicitly leverages external pre-trained generative models for implicit mutations and separate multi-label classifiers for objectives, without any self-definitional loops, fitted parameters renamed as predictions, or load-bearing self-citations in the provided text. The central proposal is a methodological instantiation that remains independent of its own outputs or prior author results.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters or invented entities; the central approach rests on the domain assumption that classifier label predictions can serve as effective optimization objectives for user preferences.

axioms (1)

domain assumption The labels predicted by pre-trained multi-label image classifiers can be used as effective objectives for optimizing image generation to match user preferences
This is invoked when the abstract states that predicted labels serve as multiple objectives to optimize.

pith-pipeline@v0.9.0 · 5713 in / 1356 out tokens · 45291 ms · 2026-05-24T08:57:38.640760+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages · 5 internal anchors

[1]

Crowson, Katherine, Stella Biderman, Daniel Kornis, Dashiell Stander, Eric Hallahan, Louis Castricato, and Edward Raff. ”Vqgan-clip: Open domain image generation and editing with natural language guidance.” In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XXXVII, pp. 88-105. Cham: Springer Na...

work page 2022
[2]

”Text-guided synthesis of artistic images with retrieval-augmented diffusion models.” arXiv preprint arXiv:2207.13038 (2022)

Rombach, Robin, Andreas Blattmann, and Bj ¨orn Ommer. ”Text-guided synthesis of artistic images with retrieval-augmented diffusion models.” arXiv preprint arXiv:2207.13038 (2022)

work page arXiv 2022
[3]

Chen, Wenhu, Hexiang Hu, Chitwan Saharia, and William W. Co- hen. ”Re-imagen: Retrieval-augmented text-to-image generator.” arXiv preprint arXiv:2209.14491 (2022)

work page arXiv 2022
[4]

Wang, Yunlong, Shuyuan Shen, and Brian Y . Lim. ”RePrompt: Au- tomatic Prompt Editing to Refine AI-Generative Art Towards Precise Expressions.” arXiv preprint arXiv:2302.09466 (2023)

work page arXiv 2023
[5]

The family that plays together stays together

Oppenlaender, Jonas. ”A Taxonomy of Prompt Modifiers for Text-to- Image Generation.” arXiv preprint arXiv:2204.13988 (2022). 0.5 0.6 0.7 0.8 0.9 1.0 Objective 2: child riding bicycle (higher is better) 0.5 0.6 0.7 0.8 0.9 1.0 Objective 1: person helping child (higher is better) Before evolution, most generated images did not satisfy objective 1. After evo...

work page arXiv 2022
[6]

Liu, Vivian, and Lydia B. Chilton. ”Design guidelines for prompt engineering text-to-image generative models.” In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems, pp. 1-23. 2022

work page 2022
[7]

”Grimm in Wonderland: Prompt Engineering with Midjourney to Illustrate Fairytales.” arXiv preprint arXiv:2302.08961 (2023)

Ruskov, Martin. ”Grimm in Wonderland: Prompt Engineering with Midjourney to Illustrate Fairytales.” arXiv preprint arXiv:2302.08961 (2023)

work page arXiv 2023
[8]

eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers

Balaji, Yogesh, Seungjun Nah, Xun Huang, Arash Vahdat, Jiaming Song, Karsten Kreis, Miika Aittala et al. ”ediffi: Text-to-image dif- fusion models with an ensemble of expert denoisers.” arXiv preprint arXiv:2211.01324 (2022)

work page internal anchor Pith review Pith/arXiv arXiv 2022
[9]

”High-resolution image synthesis with latent diffu- sion models.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Rombach, Robin, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. ”High-resolution image synthesis with latent diffu- sion models.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10684-10695. 2022

work page 2022
[10]

Hierarchical Text-Conditional Image Generation with CLIP Latents

Ramesh, Aditya, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. ”Hierarchical text-conditional image generation with clip latents.” arXiv preprint arXiv:2204.06125 (2022)

work page internal anchor Pith review Pith/arXiv arXiv 2022
[11]

Tian, Yingtao, and David Ha. ”Modern evolution strategies for creativity: Fitting concrete images and abstract concepts.” In Artificial Intelligence in Music, Sound, Art and Design: 11th International Conference, Evo- MUSART 2022, Held as Part of EvoStar 2022, Madrid, Spain, April 20–22, 2022, Proceedings, pp. 275-291. Cham: Springer International Publish...

work page 2022
[12]

”Many-Objective Optimization for Diverse Image Generation.” (2021)

Rakotonirina, Nathana ¨el Carraz, Andry Rasoanaivo, Laurent Najman, Petr Kungurtsev, Jeremy Rapin, Fabien Teytaud, Baptiste Roziere et al. ”Many-Objective Optimization for Diverse Image Generation.” (2021)

work page 2021
[13]

Radford, Alec, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry et al. 2021. ”Learning transferable visual models from natural language supervision.” PMLR. 8748-8763

work page 2021
[14]

Achlioptas, Panos, Maks Ovsjanikov, Kilichbek Haydarov, Mohamed Elhoseiny, and Leonidas J. Guibas. ”Artemis: Affective language for visual art.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11569-11579. 2021

work page 2021
[15]

”Learning transferable visual models from natural language supervision.” In International conference on machine learning, pp

Radford, Alec, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry et al. ”Learning transferable visual models from natural language supervision.” In International conference on machine learning, pp. 8748-8763. PMLR, 2021

work page 2021
[16]

”Tresnet: High performance gpu-dedicated architecture.” In proceedings of the IEEE/CVF winter conference on applications of computer vision, pp

Ridnik, Tal, Hussam Lawen, Asaf Noy, Emanuel Ben Baruch, Gilad Sharir, and Itamar Friedman. ”Tresnet: High performance gpu-dedicated architecture.” In proceedings of the IEEE/CVF winter conference on applications of computer vision, pp. 1400-1409. 2021

work page 2021
[17]

An Explanation of In-context Learning as Implicit Bayesian Inference

Xie, Sang Michael, Aditi Raghunathan, Percy Liang, and Tengyu Ma. ”An explanation of in-context learning as implicit bayesian inference.” arXiv preprint arXiv:2111.02080 (2021)

work page internal anchor Pith review Pith/arXiv arXiv 2021
[18]

The Power of Scale for Parameter-Efficient Prompt Tuning

Lester, Brian, Rami Al-Rfou, and Noah Constant. ”The power of scale for parameter-efficient prompt tuning.” arXiv preprint arXiv:2104.08691 (2021)

work page internal anchor Pith review Pith/arXiv arXiv 2021
[19]

”Fine- tuning language models to find agreement among humans with diverse preferences.” Advances in Neural Information Processing Systems 35 (2022): 38176-38189

Bakker, Michiel, Martin Chadwick, Hannah Sheahan, Michael Tessler, Lucy Campbell-Gillingham, Jan Balaguer, Nat McAleese et al. ”Fine- tuning language models to find agreement among humans with diverse preferences.” Advances in Neural Information Processing Systems 35 (2022): 38176-38189

work page 2022
[20]

Evolutionary Multitasking for Multiobjective Continuous Optimization: Benchmark Problems, Performance Metrics and Baseline Results

Yuan, Yuan, Yew-Soon Ong, Liang Feng, A. Kai Qin, Abhishek Gupta, Bingshui Da, Qingfu Zhang, Kay Chen Tan, Yaochu Jin, and Hisao Ishibuchi. ”Evolutionary multitasking for multiobjective continuous op- timization: Benchmark problems, performance metrics and baseline results.” arXiv preprint arXiv:1706.02766 (2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017
[21]

Deb, Kalyanmoy, Amrit Pratap, Sameer Agarwal, and T. A. M. T. Meyarivan. ”A fast and elitist multiobjective genetic algorithm: NSGA- II.” IEEE transactions on evolutionary computation 6, no. 2 (2002): 182- 197

work page 2002
[22]

”Back to the roots: Multi-x evolutionary computation.” Cognitive Computation 11 (2019): 1-17

Gupta, Abhishek, and Yew-Soon Ong. ”Back to the roots: Multi-x evolutionary computation.” Cognitive Computation 11 (2019): 1-17

work page 2019

[1] [1]

Crowson, Katherine, Stella Biderman, Daniel Kornis, Dashiell Stander, Eric Hallahan, Louis Castricato, and Edward Raff. ”Vqgan-clip: Open domain image generation and editing with natural language guidance.” In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XXXVII, pp. 88-105. Cham: Springer Na...

work page 2022

[2] [2]

”Text-guided synthesis of artistic images with retrieval-augmented diffusion models.” arXiv preprint arXiv:2207.13038 (2022)

Rombach, Robin, Andreas Blattmann, and Bj ¨orn Ommer. ”Text-guided synthesis of artistic images with retrieval-augmented diffusion models.” arXiv preprint arXiv:2207.13038 (2022)

work page arXiv 2022

[3] [3]

Chen, Wenhu, Hexiang Hu, Chitwan Saharia, and William W. Co- hen. ”Re-imagen: Retrieval-augmented text-to-image generator.” arXiv preprint arXiv:2209.14491 (2022)

work page arXiv 2022

[4] [4]

Wang, Yunlong, Shuyuan Shen, and Brian Y . Lim. ”RePrompt: Au- tomatic Prompt Editing to Refine AI-Generative Art Towards Precise Expressions.” arXiv preprint arXiv:2302.09466 (2023)

work page arXiv 2023

[5] [5]

The family that plays together stays together

Oppenlaender, Jonas. ”A Taxonomy of Prompt Modifiers for Text-to- Image Generation.” arXiv preprint arXiv:2204.13988 (2022). 0.5 0.6 0.7 0.8 0.9 1.0 Objective 2: child riding bicycle (higher is better) 0.5 0.6 0.7 0.8 0.9 1.0 Objective 1: person helping child (higher is better) Before evolution, most generated images did not satisfy objective 1. After evo...

work page arXiv 2022

[6] [6]

Liu, Vivian, and Lydia B. Chilton. ”Design guidelines for prompt engineering text-to-image generative models.” In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems, pp. 1-23. 2022

work page 2022

[7] [7]

”Grimm in Wonderland: Prompt Engineering with Midjourney to Illustrate Fairytales.” arXiv preprint arXiv:2302.08961 (2023)

Ruskov, Martin. ”Grimm in Wonderland: Prompt Engineering with Midjourney to Illustrate Fairytales.” arXiv preprint arXiv:2302.08961 (2023)

work page arXiv 2023

[8] [8]

eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers

Balaji, Yogesh, Seungjun Nah, Xun Huang, Arash Vahdat, Jiaming Song, Karsten Kreis, Miika Aittala et al. ”ediffi: Text-to-image dif- fusion models with an ensemble of expert denoisers.” arXiv preprint arXiv:2211.01324 (2022)

work page internal anchor Pith review Pith/arXiv arXiv 2022

[9] [9]

”High-resolution image synthesis with latent diffu- sion models.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Rombach, Robin, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. ”High-resolution image synthesis with latent diffu- sion models.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10684-10695. 2022

work page 2022

[10] [10]

Hierarchical Text-Conditional Image Generation with CLIP Latents

Ramesh, Aditya, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. ”Hierarchical text-conditional image generation with clip latents.” arXiv preprint arXiv:2204.06125 (2022)

work page internal anchor Pith review Pith/arXiv arXiv 2022

[11] [11]

Tian, Yingtao, and David Ha. ”Modern evolution strategies for creativity: Fitting concrete images and abstract concepts.” In Artificial Intelligence in Music, Sound, Art and Design: 11th International Conference, Evo- MUSART 2022, Held as Part of EvoStar 2022, Madrid, Spain, April 20–22, 2022, Proceedings, pp. 275-291. Cham: Springer International Publish...

work page 2022

[12] [12]

”Many-Objective Optimization for Diverse Image Generation.” (2021)

Rakotonirina, Nathana ¨el Carraz, Andry Rasoanaivo, Laurent Najman, Petr Kungurtsev, Jeremy Rapin, Fabien Teytaud, Baptiste Roziere et al. ”Many-Objective Optimization for Diverse Image Generation.” (2021)

work page 2021

[13] [13]

Radford, Alec, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry et al. 2021. ”Learning transferable visual models from natural language supervision.” PMLR. 8748-8763

work page 2021

[14] [14]

Achlioptas, Panos, Maks Ovsjanikov, Kilichbek Haydarov, Mohamed Elhoseiny, and Leonidas J. Guibas. ”Artemis: Affective language for visual art.” In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11569-11579. 2021

work page 2021

[15] [15]

”Learning transferable visual models from natural language supervision.” In International conference on machine learning, pp

Radford, Alec, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry et al. ”Learning transferable visual models from natural language supervision.” In International conference on machine learning, pp. 8748-8763. PMLR, 2021

work page 2021

[16] [16]

”Tresnet: High performance gpu-dedicated architecture.” In proceedings of the IEEE/CVF winter conference on applications of computer vision, pp

Ridnik, Tal, Hussam Lawen, Asaf Noy, Emanuel Ben Baruch, Gilad Sharir, and Itamar Friedman. ”Tresnet: High performance gpu-dedicated architecture.” In proceedings of the IEEE/CVF winter conference on applications of computer vision, pp. 1400-1409. 2021

work page 2021

[17] [17]

An Explanation of In-context Learning as Implicit Bayesian Inference

Xie, Sang Michael, Aditi Raghunathan, Percy Liang, and Tengyu Ma. ”An explanation of in-context learning as implicit bayesian inference.” arXiv preprint arXiv:2111.02080 (2021)

work page internal anchor Pith review Pith/arXiv arXiv 2021

[18] [18]

The Power of Scale for Parameter-Efficient Prompt Tuning

Lester, Brian, Rami Al-Rfou, and Noah Constant. ”The power of scale for parameter-efficient prompt tuning.” arXiv preprint arXiv:2104.08691 (2021)

work page internal anchor Pith review Pith/arXiv arXiv 2021

[19] [19]

”Fine- tuning language models to find agreement among humans with diverse preferences.” Advances in Neural Information Processing Systems 35 (2022): 38176-38189

Bakker, Michiel, Martin Chadwick, Hannah Sheahan, Michael Tessler, Lucy Campbell-Gillingham, Jan Balaguer, Nat McAleese et al. ”Fine- tuning language models to find agreement among humans with diverse preferences.” Advances in Neural Information Processing Systems 35 (2022): 38176-38189

work page 2022

[20] [20]

Evolutionary Multitasking for Multiobjective Continuous Optimization: Benchmark Problems, Performance Metrics and Baseline Results

Yuan, Yuan, Yew-Soon Ong, Liang Feng, A. Kai Qin, Abhishek Gupta, Bingshui Da, Qingfu Zhang, Kay Chen Tan, Yaochu Jin, and Hisao Ishibuchi. ”Evolutionary multitasking for multiobjective continuous op- timization: Benchmark problems, performance metrics and baseline results.” arXiv preprint arXiv:1706.02766 (2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017

[21] [21]

Deb, Kalyanmoy, Amrit Pratap, Sameer Agarwal, and T. A. M. T. Meyarivan. ”A fast and elitist multiobjective genetic algorithm: NSGA- II.” IEEE transactions on evolutionary computation 6, no. 2 (2002): 182- 197

work page 2002

[22] [22]

”Back to the roots: Multi-x evolutionary computation.” Cognitive Computation 11 (2019): 1-17

Gupta, Abhishek, and Yew-Soon Ong. ”Back to the roots: Multi-x evolutionary computation.” Cognitive Computation 11 (2019): 1-17

work page 2019