ImageDDI: Image-enhanced Molecular Motif Sequence Representation for Drug-Drug Interaction Prediction

Bosheng Song; Chaoyi Li; Hongxin Xiang; Jianmin Wang; Pengsen Ma; Tengfei Ma; Xiangxiang Zeng; Yiping Liu; Yuqin He

REVIEW 4 major objections 2 minor 15 references

ImageDDI: Image-enhanced Molecular Motif Sequence Representation for Drug-Drug Interaction Prediction

T0 review · 4 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read ImageDDI claims that reading a drug pair as one sequence of functional motifs, enhanced by adaptively fused molecular image features, outperforms state-of-the-art DDI prediction on standard benchmarks.

desk verdict The supplied full text is not ImageDDI — it is an unrelated X-ray diffraction paper — so the abstract's DDI claims are unverifiable and this submission should not be reviewed as-is. read the letter →

arxiv 2508.08338 v1 pith:X7FFX32M submitted 2025-08-11 cs.CV cs.AIcs.LG

Yuqin He , Tengfei Ma , Chaoyi Li , Pengsen Ma , Hongxin Xiang , Jianmin Wang , Yiping Liu , Bosheng Song

show 1 more author

Xiangxiang Zeng

This is my paper · ORCID

classification cs.CVcs.AIcs.LG

keywords drug-druginteractionpredictionfunctionalmotifstransformerencodermolecularimagerepresentationadaptivefeaturefusionpolypharmacydeeplearningmolecule

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

ImageDDI sets out to show that drug-drug interactions are best predicted from the local sites where molecules actually touch, rather than from whole-molecule structure. It cuts each drug into functional motifs, strings the two drugs' motifs into a single sequence, and encodes that sequence with a transformer. A second stream renders each molecule as an image and captures global cues such as texture, shadow, color, and planar spatial arrangement; an Adaptive Feature Fusion module decides how much of this visual information to blend into the motif sequence. The paper claims that this combined representation outperforms state-of-the-art DDI methods on widely used datasets and remains competitive when the image stream is 2D or 3D.

What carries the argument

Functional-motif tokenization plus pair-level sequence encoding: each drug is decomposed into functional motifs and the two motif lists are merged into one order-sensitive token sequence, letting the transformer see both drugs at once. Molecular image encoding supplies global spatial cues such as texture, shadow, color, and planar relationships. Adaptive Feature Fusion is the third component: instead of fixed concatenation, it dynamically weights how much image information is blended with the motif-sequence representation for each input.

What would settle it

Run ImageDDI on the public DDI benchmark datasets with the same train/test split used in the paper, and compare against (a) the transformer over motif sequences alone with no image stream and (b) a variant with fixed, non-adaptive fusion. If either matches or beats ImageDDI, the claimed advantage of the adaptive image stream is not established. As a second check, changing only the molecular image rendering style should not materially change accuracy if the fusion is learning chemistry rather than visual artifacts.

Watch

Extended reading notes

Core claim

Drug-drug interactions are fundamentally motif interactions, so the paper's central claim is that representing a drug pair as a single sequence of functional motifs, then enhancing that local representation with global molecular image information, is what improves prediction. ImageDDI tokenizes each drug into functional motifs, concatenates the two drugs' motif sequences, and feeds the joint sequence to a transformer encoder. It also encodes molecular images to capture global spatial features, then uses Adaptive Feature Fusion to combine the two streams. On standard DDI benchmarks the paper reports that this fused model outperforms state-of-the-art approaches and performs competitively in bo

Load-bearing premise

That segmenting each drug into functional motifs, merging the pair's motifs into one transformer-readable sequence, and adaptively fusing molecule-image features yields a representation that generalizes to unseen drug pairs; the available abstract supplies no rule for the segmentation, no rendering or embedding details, and no validation.

Editorial extensions

If this is right

If the claim holds, DDI prediction on standard benchmarks improves, helping flag adverse interactions in multi-drug regimens before prescription.
The pair-as-one-sequence design gives the transformer direct access to cross-drug motif adjacencies, which may expose interaction mechanisms that per-drug encoders miss.
Adaptive Feature Fusion suggests the visual and sequence modalities are complementary and that fixed-ratio fusion is suboptimal, a design choice that can transfer to other molecular pair-prediction tasks.
Competitive results in both 2D and 3D image-enhanced scenarios imply the method does not depend on a single particular rendering of molecular images.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the supplied material contains no ablation study, the paper's own claim leaves open whether the gain comes from the joint motif sequence, the image stream, or the fusion rule; testing the three variants on identical splits would settle it.
The same pair-as-one-sequence-plus-image recipe transfers naturally to drug-target interaction, drug synergy, or adverse-event prediction, where local substructures and global molecular shape both matter.
If the visual stream is truly complementary, then molecular image rendering choices—such as resolution, style, and 2D versus 3D depiction—become hyperparameters that should be reported and standardized for fair comparison.
The supplied full text belongs to a different manuscript, so the claims here rest on the abstract alone; implementation details such as the motif segmentation rule and the image embedding are not verifiable from the provided material.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick note on arXiv:2508.08338. The abstract describes ImageDDI, a drug-drug interaction predictor that tokenizes molecules into functional motifs, concatenates a drug pair's motifs into a single transformer-readable sequence, and fuses in global molecular image features via adaptive fusion. That is a reasonable design and the abstract is coherent and clearly written. Nothing in the supplied full text, however, matches it. The body is DiffractGPT, a separate manuscript about crystal structure determination from powder X-ray patterns. There are no ImageDDI methods, equations, datasets, tables, baselines, or ablations anywhere in the submission. The sentence in the abstract claiming ImageDDI outperforms state-of-the-art methods is backed by zero quantitative evidence in the supplied material. This is not a weak ablation or a thin comparison; the contribution itself is absent. What is present is a plausible abstract plus an unrelated manuscript. So I cannot assess soundness, reproducibility, or even whether the proposed architecture exists beyond the abstract's description. The underlying idea — combining functional-motif sequence encoding with image-based spatial features — is not implausible, and the abstract implies the authors have run experiments, but no experiment is visible. That said, the mismatch is likely a submission error rather than evidence of fraud: DiffractGPT reads like a legitimate, self-contained preprint. Still, as submitted, the work is internally incoherent: the title, abstract, and metadata all describe ImageDDI, while the body describes something else entirely. My recommendation: desk reject this version and ask the authors to upload the correct manuscript. If the actual ImageDDI paper delivers what the abstract promises — standard DDI datasets, baselines, ablations, error bars — it is worth a proper peer review. As it stands, there is nothing to referee. I would not cite it or bring it to reading group in this form.

Referee Report

4 major / 2 minor

Summary. The submission's abstract describes ImageDDI, a framework that represents drug pairs via functional-motif sequences and molecular images, with a transformer encoder and Adaptive Feature Fusion, and claims it outperforms state-of-the-art methods on widely used drug-drug interaction (DDI) datasets. The abstract is the only part of the manuscript that concerns ImageDDI; the supplied full text is an unrelated manuscript, DiffractGPT (arXiv:2508.08349v1), which addresses atomic structure determination from X-ray diffraction patterns. The full text contains no description of ImageDDI, no DDI datasets, no motif tokenization details, no fusion mechanism, no experimental results, and no comparison to prior DDI methods. Consequently, the paper as supplied provides no verifiable methods or evidence for its central claims.

Significance. If the claims in the abstract were fully supported by a complete manuscript, ImageDDI would represent a plausible contribution to multimodal molecular representation learning for DDI prediction, with the novel element of adaptively fusing global molecular image features with a local motif-sequence transformer. However, the supplied material contains none of the architecture, implementation, or evaluation needed to substantiate this. There are no machine-checked proofs, no reproducible code, no parameter-free derivations, and no falsifiable predictions presented. The significance of the work therefore cannot be assessed from the submitted manuscript; at present the only assessable claim is contradicted by the absence of any supporting content.

major comments (4)

[Full Text (entire supplied manuscript)] The full text is a completely different paper, DiffractGPT, concerned with X-ray diffraction and crystal structures. It never mentions ImageDDI, drug-drug interactions, functional motifs, molecular images, or Adaptive Feature Fusion. The central assertion of the abstract—that ImageDDI outperforms state-of-the-art DDI methods—therefore has no corresponding methods or results anywhere in the manuscript. This is not a local technical gap but a total absence of the proposed model and its evaluation, making the central claim unsupportable as submitted.
[Abstract (method description)] The abstract introduces several load-bearing components without definition: 'functional motifs,' 'combines their motifs into a single sequence,' 'transformer-based encoder,' and 'Adaptive Feature Fusion.' None of these are specified at the level required to understand, implement, or verify the method. In particular, there is no description of how motifs are segmented, how the single sequence is constructed, what image representations are used, or how the fusion is made adaptive. Without these details, the manuscript's claimed contribution is not reproducible.
[Abstract (experimental claims)] The statement that ImageDDI 'outperforms state-of-the-art methods' is not supported by any table, figure, dataset name, baseline, metric, error bar, or statistical test. The abstract also mentions '2D and 3D image-enhanced scenarios' without defining what these scenarios are. Since the full text provides no experimental section for ImageDDI, the empirical claims are entirely unverifiable.
[Full Text (reproducibility)] No code, data, model weights, or training details are provided for ImageDDI. The only code references in the full text are for DiffractGPT (e.g., the AtomGPT GitHub page), which is a different system. Even if the abstract's method were plausible, the absence of any implementation details or evaluation protocol would preclude reproducibility.

minor comments (2)

[Abstract] The phrase 'widely used datasets' should name the specific DDI benchmarks (e.g., DrugBank, Twosides, ZhangDDI) and report their statistics; as written, the claim is untestable.
[Abstract] The term 'Adaptive Feature Fusion' is used without a citation or comparison to prior fusion mechanisms, making it difficult to situate the claimed novelty.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation chain is present because the supplied full text is an unrelated manuscript; the claimed ImageDDI method and experiments are absent, so there is no equation-level reduction to audit.

full rationale

The abstract of arXiv:2508.08338 describes ImageDDI, a supervised deep-learning framework for drug-drug interaction prediction that tokenizes molecules into functional motifs, encodes a concatenated motif sequence with a transformer, and fuses global molecular image features via Adaptive Feature Fusion. However, the supplied full text is arXiv:2508.08349v1, 'DiffractGPT: Atomic Structure Determination from X-ray Diffraction Patterns using Generative Pre-trained Transformer' by Kamal Choudhary, which contains no description of ImageDDI, no DDI datasets, no motif tokenization rule, no image rendering/embedding details, no fusion equations, and no experimental results. Because none of the proposed method's components are defined in the supplied material, there is no equation-level derivation chain to audit: no parameter is fitted to data and then renamed as a prediction, no quantity is defined in terms of another in a circular way, and no load-bearing self-citation chain is invoked. The abstract's claim that ImageDDI 'outperforms state-of-the-art methods' is therefore unverifiable from the provided text, but unsupportedness is a completeness/integrity issue, not a circularity issue. Per the instruction to claim circularity only when the specific reduction can be quoted, no circular step can be exhibited, so the circularity score is 0. The mismatch between the abstract and the body should be raised as a correctness/integrity concern, not as circular reasoning.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The ledger is necessarily partial because only the abstract was reviewable. No explicit free parameters are disclosed; learned network weights (including the Adaptive Feature Fusion gates) are trained, not fitted constants with reported values. No new physical entities are introduced; 'functional motifs' are an existing representational concept, not a newly postulated object.

assumptions (3)

domain assumption DDIs are fundamentally caused by motif interactions rather than overall drug structures.
Stated as the motivating premise in the abstract; the entire motif-sequence representation is designed around it, but no biochemical evidence is cited in the abstract.
domain assumption Global molecular image information (texture, shadow, color, planar spatial relationships) captures structure complementary to the motif sequence and improves prediction.
The abstract asserts image enhancement 'further enhances the spatial representation'; this transfer of visual features to molecular function is assumed, not derived.
domain assumption Reported gains on 'widely used datasets' with standard splits are meaningful for general DDI prediction.
The evaluation claim rests on benchmark datasets and SOTA comparisons; dataset details, splits, and leakage checks are absent from the abstract and from the mismatched body.

how reviews work

0 comments

Cite this review

Pith. "Pith review of ImageDDI: Image-enhanced Molecular Motif Sequence Representation for Drug-Drug Interaction Prediction." pith.science (2026). https://pith.science/paper/X7FFX32M

@misc{pith2026250808338,
  author       = {Pith},
  title        = {Pith review of: ImageDDI: Image-enhanced Molecular Motif Sequence Representation for Drug-Drug Interaction Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X7FFX32M}},
  note         = {Machine review of arXiv:2508.08338}
}

read the original abstract

To mitigate the potential adverse health effects of simultaneous multi-drug use, including unexpected side effects and interactions, accurately identifying and predicting drug-drug interactions (DDIs) is considered a crucial task in the field of deep learning. Although existing methods have demonstrated promising performance, they suffer from the bottleneck of limited functional motif-based representation learning, as DDIs are fundamentally caused by motif interactions rather than the overall drug structures. In this paper, we propose an Image-enhanced molecular motif sequence representation framework for \textbf{DDI} prediction, called ImageDDI, which represents a pair of drugs from both global and local structures. Specifically, ImageDDI tokenizes molecules into functional motifs. To effectively represent a drug pair, their motifs are combined into a single sequence and embedded using a transformer-based encoder, starting from the local structure representation. By leveraging the associations between drug pairs, ImageDDI further enhances the spatial representation of molecules using global molecular image information (e.g. texture, shadow, color, and planar spatial relationships). To integrate molecular visual information into functional motif sequence, ImageDDI employs Adaptive Feature Fusion, enhancing the generalization of ImageDDI by dynamically adapting the fusion process of feature representations. Experimental results on widely used datasets demonstrate that ImageDDI outperforms state-of-the-art methods. Moreover, extensive experiments show that ImageDDI achieved competitive performance in both 2D and 3D image-enhanced scenarios compared to other models.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 12 canonical work pages

[8]

M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M

(24) Maffettone, P. M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M. A.; Olds, D.; Lud- wig, A.; Cooper, A. I. Crystallography companion agent for high-throughput materials discovery. Nature Computational Science 2021, 1, 290–297. (25) Oviedo, F.; Ren, Z.; Sun, S.; Settens, C.; Liu, Z.; Hartono, N. T. P.; Ramasamy, S.; DeCost, B. L.; Tian, S. I.; Roman...

work page 2021
[59]

K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S

(12) Vasudevan, R. K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S. V.; Hattrick-Simpers, J. Materials science in the artificial intelligence age: high-throughput library generation, machine learning, and a pathway from correlations to the underpinning physics. MRS communications 2019, 9, 821–838. (13)...

work page 2019
[60]

Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning

(26) Chen, L.; Wang, B.; Zhang, W.; Zheng, S.; Chen, Z.; Zhang, M.; Dong, C.; Pan, F.; Li, S. Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning. Journal of the American Chemical Society 2024, 146,

work page 2024
[83]

https://www.nist.gov/chips, [Accessed 10-10-2024]

21 (14) CHIPS.Gov — nist.gov. https://www.nist.gov/chips, [Accessed 10-10-2024]. (15) Surdu, V.-A.; Gy˝ orgy, R. X-ray diffraction data analysis by machine learning meth- ods—a review. Applied Sciences 2023, 13,

work page 2024
[86]

GenX 3: the latest generation of an established tool

(8) Glavic, A.; Bj¨ orck, M. GenX 3: the latest generation of an established tool. Journal of applied crystallography 2022, 55, 1063–1071. (9) Coelho, A. A. TOPAS and TOPAS-Academic: an optimization program integrating computer algebra and crystallographic objects written in C++. Journal of Applied Crystallography 2018, 51, 210–218. (10) Wenk, H.-R.; Lutt...

work page 2022
[104]

Artifact identifica- tion in X-ray diffraction data using machine learning methods

(20) Yanxon, H.; Weng, J.; Parraga, H.; Xu, W.; Ruett, U.; Schwarz, N. Artifact identifica- tion in X-ray diffraction data using machine learning methods. Journal of Synchrotron Radiation 2023, 30, 137–146. (21) Zaloga, A. N.; Stanovov, V. V.; Bezrukova, O. E.; Dubinin, P. S.; Yakimov, I. S. Crystal symmetry classification from powder X-ray diffraction pa...

work page 2023
[115]

Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials

(53) Boulle, A.; Debelle, A. Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials. Machine Learning: Science and Technology 2023, 4, 015002. 26 (54) Judge, W.; Chan, H.; Sankaranarayanan, S.; Harder, R. J.; Cabana, J.; Cherukara, M. J. Defect identification in simulated Bragg coherent dif...

work page 2023
[173]

Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D

(39) Jiang, A. Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D. S.; Casas, D. d. l.; Bressand, F.; Lengyel, G.; Lample, G.; Saulnier, L.; others Mistral 7B. arXiv preprint arXiv:2310.06825 2023, [Accessed 10-10-2024]. (40) Jumper, J.; Evans, R.; Pritzel, A.; Green, T.; Figurnov, M.; Ronneberger, O.; Tunya- suvunakool, K.; Bates, R.; ˇZ ´ ıdek, A...

arXiv 2023

Show all 15 references

[1569]

F.; DeCost, B.; Biacchi, A

(37) Wines, D.; Gurunathan, R.; Garrity, K. F.; DeCost, B.; Biacchi, A. J.; Tavazza, F.; Choudhary, K. Recent progress in the JAR VIS infrastructure for next-generation data- driven materials design. Applied Physics Reviews 2023, 10 . 24 (38) Choudhary, K.; Garrity, K. F.; Rei...

2023
[2002]

X-ray metrology for advanced microelectronics

(3) Wyon, C. X-ray metrology for advanced microelectronics. The European Physical Journal-Applied Physics 2010, 49, 20101. (4) Holder, C. F.; Schaak, R. E. Tutorial on powder X-ray diffraction for characterizing nanoscale materials. ACS Nano 2019, 13, 7359–7365. 20 (5) Brown, ...

2010
[2012]

FullProf

(6) Rodriguez-Carvajal, J.; Roisnel, T. FullProf. 98 and WinPLOTR: new windows 95/NT applications for diffraction. Commission for Powder Diffraction, International Union of Crystallography, Newsletter 1998, 20, May–August. (7) Larson, A. C.; Von Dreele, R. B. General Structure...

1998
[2021]

A brief overview of ChatGPT: The history, status quo and potential future development

(33) Wu, T.; He, S.; Liu, J.; Sun, S.; Liu, K.; Han, Q.-L.; Tang, Y. A brief overview of ChatGPT: The history, status quo and potential future development. IEEE/CAA Journal of Automatica Sinica 2023, 10, 1122–1136. (34) Pimentel, A.; Wagener, A.; da Silveira, E. F.; Picciani, ...

2023 doi
[2024]

M.; Schwaller, P.; Ortega-Guerrero, A.; Smit, B

(35) Jablonka, K. M.; Schwaller, P.; Ortega-Guerrero, A.; Smit, B. Leveraging large language models for predictive chemistry. Nature Machine Intelligence 2024, 1–9. (36) Polak, M. P.; Morgan, D. Extracting accurate materials data from research papers with conversational langua...

2024
[8098]

Machine learning-accelerated discovery of novel 2D ferromagnetic materials with strong magnetization

(27) Xin, C.; Yin, Y.; Song, B.; Fan, Z.; Song, Y.; Pan, F. Machine learning-accelerated discovery of novel 2D ferromagnetic materials with strong magnetization. Chip 2023, 2, 100071. (28) Xin, C.; Song, B.; Jin, G.; Song, Y.; Pan, F. Advancements in High-Throughput Screen- in...

2023
[9992]

B.; Chung, J.; Jung, J.; Sohn, K.; Singh, S

(16) Park, W. B.; Chung, J.; Jung, J.; Sohn, K.; Singh, S. P.; Pyo, M.; Shin, N.; Sohn, K.-S. Classification of crystal structure using a convolutional neural network. IUCrJ 2017, 4, 486–494. (17) Zhdanov, M.; Zhdanov, A. Machine learning-assisted close-set X-ray diffraction p...

2017 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [8]

M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M

(24) Maffettone, P. M.; Banko, L.; Cui, P.; Lysogorskiy, Y.; Little, M. A.; Olds, D.; Lud- wig, A.; Cooper, A. I. Crystallography companion agent for high-throughput materials discovery. Nature Computational Science 2021, 1, 290–297. (25) Oviedo, F.; Ren, Z.; Sun, S.; Settens, C.; Liu, Z.; Hartono, N. T. P.; Ramasamy, S.; DeCost, B. L.; Tian, S. I.; Roman...

work page 2021

[2] [59]

K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S

(12) Vasudevan, R. K.; Choudhary, K.; Mehta, A.; Smith, R.; Kusne, G.; Tavazza, F.; Vlcek, L.; Ziatdinov, M.; Kalinin, S. V.; Hattrick-Simpers, J. Materials science in the artificial intelligence age: high-throughput library generation, machine learning, and a pathway from correlations to the underpinning physics. MRS communications 2019, 9, 821–838. (13)...

work page 2019

[3] [60]

Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning

(26) Chen, L.; Wang, B.; Zhang, W.; Zheng, S.; Chen, Z.; Zhang, M.; Dong, C.; Pan, F.; Li, S. Crystal Structure Assignment for Unknown Compounds from X-ray Diffraction Patterns with Deep Learning. Journal of the American Chemical Society 2024, 146,

work page 2024

[4] [83]

https://www.nist.gov/chips, [Accessed 10-10-2024]

21 (14) CHIPS.Gov — nist.gov. https://www.nist.gov/chips, [Accessed 10-10-2024]. (15) Surdu, V.-A.; Gy˝ orgy, R. X-ray diffraction data analysis by machine learning meth- ods—a review. Applied Sciences 2023, 13,

work page 2024

[5] [86]

GenX 3: the latest generation of an established tool

(8) Glavic, A.; Bj¨ orck, M. GenX 3: the latest generation of an established tool. Journal of applied crystallography 2022, 55, 1063–1071. (9) Coelho, A. A. TOPAS and TOPAS-Academic: an optimization program integrating computer algebra and crystallographic objects written in C++. Journal of Applied Crystallography 2018, 51, 210–218. (10) Wenk, H.-R.; Lutt...

work page 2022

[6] [104]

Artifact identifica- tion in X-ray diffraction data using machine learning methods

(20) Yanxon, H.; Weng, J.; Parraga, H.; Xu, W.; Ruett, U.; Schwarz, N. Artifact identifica- tion in X-ray diffraction data using machine learning methods. Journal of Synchrotron Radiation 2023, 30, 137–146. (21) Zaloga, A. N.; Stanovov, V. V.; Bezrukova, O. E.; Dubinin, P. S.; Yakimov, I. S. Crystal symmetry classification from powder X-ray diffraction pa...

work page 2023

[7] [115]

Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials

(53) Boulle, A.; Debelle, A. Convolutional neural network analysis of x-ray diffraction data: strain profile retrieval in ion beam modified materials. Machine Learning: Science and Technology 2023, 4, 015002. 26 (54) Judge, W.; Chan, H.; Sankaranarayanan, S.; Harder, R. J.; Cabana, J.; Cherukara, M. J. Defect identification in simulated Bragg coherent dif...

work page 2023

[8] [173]

Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D

(39) Jiang, A. Q.; Sablayrolles, A.; Mensch, A.; Bamford, C.; Chaplot, D. S.; Casas, D. d. l.; Bressand, F.; Lengyel, G.; Lample, G.; Saulnier, L.; others Mistral 7B. arXiv preprint arXiv:2310.06825 2023, [Accessed 10-10-2024]. (40) Jumper, J.; Evans, R.; Pritzel, A.; Green, T.; Figurnov, M.; Ronneberger, O.; Tunya- suvunakool, K.; Bates, R.; ˇZ ´ ıdek, A...

arXiv 2023