Pith. sign in

REVIEW 4 major objections 4 minor 36 references

Multi-Hierarchical Fine-Grained Feature Mapping Driven by Feature Contribution for Molecular Odor Prediction

T0 review · 4 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The proposed HMFNet predicts molecular odor descriptors with F1 0.4861 and AUROC 0.9316, better than eight graph-network baselines and four feature-mapping methods on a combined odor dataset.

desk verdict Plausible architecture, but the SOTA claim rests on an unverifiable split and a CIL loss that is degenerate and doesn't implement the described co-occurrence mechanism. read the letter →

arxiv 2505.00290 v1 pith:M62U253B submitted 2025-05-01 cs.LG cs.AI

Hong Xin Xie , Jian De Sun , Fan Fu Xue , Zi Fei Han , Shan Shan Feng , Qi Chen This is my paper · ORCID
classification cs.LGcs.AI
keywords molecularodorpredictionmulti-labelclassificationgraphneuralnetworksclassimbalancefeaturecontributionharmonicmappingchemically-informedlossfingerprints
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

The paper proposes HMFNet, a multi-hierarchical network for predicting a molecule's odor descriptors from its structure, and claims it beats eight graph-network baselines and four feature-mapping methods on a combined odor dataset. The central assertion is that no single feature type is enough: atomic and bond features, global fingerprints, and SMILES-based token representations each contribute, and their contributions need to be weighted and frequency-modulated. The paper also claims that class imbalance and label co-occurrence should be handled inside the loss, via a Chemically-Informed Loss whose energy targets are adjusted by how often odor descriptors co-occur. If correct, the practical consequence is that multi-label odor prediction can be pushed to an F1 of 0.4861 and AUROC 0.9316, with most of the gain coming from the harmonic feature mapping and the loss rather than from any single representation.

What carries the argument

The machinery is a three-branch feature hierarchy plus a four-term loss. The local branch (LMFE) extracts atomic and bond features and processes them through a graph attention network; before that, a Harmonic Modulated Feature Mapping (HMFM) reweights each atomic feature by a learned importance and applies a frequency modulation $f = \sigma(\mathrm{Linear}(x'))$ against a base frequency $b = 2\pi\sigma' j/D$, producing periodic encodings $\mathrm{concat}(\cos(m\odot x'), \sin(m\odot x'))$ that are meant to capture non-smooth structure–odor relations. The global branch (GMFE) concatenates Morgan, MACCS, and Topological fingerprints with a Transformer embedding of the SMILES string. The Chemically-Informed Loss (CIL) is the weighted sum of a weighted binary cross-entropy $L_{\text{basis}}$, a chemical-property energy loss $L_{\text{class}}$ whose targets $m_{\mathrm{in}}, m_{\mathrm{out}}$ are shifted by the label co-occurrence matrix, a sample-level expected-energy loss $L_{\text{sample}}$, and a label-correlation loss $L_{\text{col}}$ that matches predicted and true label inner products. The whole system works by letting global context and label statistics reweight what the local graph encoder learns.

What would settle it

Run a faithful re-implementation with an explicit randomized train/test split, computing the co-occurrence matrix and tuning $c$, $\lambda_1$, $\lambda_2$, $\lambda_3$, $\lambda_4$ inside the training folds only. If the test F1 falls to or below the best published baseline (~0.4622 for GCast in Table 2) across seeds, or if using the all-data co-occurrence matrix in place of the training-only matrix changes F1 by a large margin, the reported state-of-the-art claim is not supported.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is an architecture-level demonstration: molecular odor prediction improves when fine-grained atomic/bond features, global graph/fingerprint/SMILES features, a learned harmonic feature mapping, and a chemically-informed multi-term loss are combined into one network. The evidence is the ablation ladder in Table 1, where F1 rises from 0.3400 (node features only) to 0.4861 (full HMFNet with CIL), and the baseline comparisons in Tables 2 and 3, where HMFM raises F1 for every base architecture it is added to and CIL raises it further. The paper notes that the best AUROC among its own configurations is 0.9356 with only node and edge features, and explains the full model's slightly lower AUROC (0.9316) as the cost of added feature streams, while arguing that the steadily increasing F1 is the better indicator for identifying true positive odor descriptors.

Load-bearing premise

The load-bearing assumption is that all label statistics used by the loss—especially the co-occurrence matrix in Equations (11)–(12)—and all tuned hyperparameters come only from training data, and that the reported F1 and AUROC are measured on a held-out test set; the paper does not state the train/test split or where those statistics are computed.

Editorial extensions

If this is right

  • If the reported comparison is correct, molecular odor prediction at scale becomes a solvable multi-label task with F1 around 0.49, making model-based screening of fragrance and flavor candidates more practical.
  • The architecture-transfer results (every one of eight graph baselines improves when HMFM is added) imply the harmonic feature mapping is a reusable module, not a fixture of one network.
  • The loss results imply that class-imbalance and label-dependency information, especially descriptor co-occurrence statistics, should be encoded explicitly in the objective for imbalanced molecular property prediction.
  • The full model trades a small AUROC decrease for a large F1 increase, so applications should choose the feature set and loss by whether they care more about ranking or about positive-label recall.
  • Combining multiple descriptor families (fingerprints, SMILES tokens, graph topology) is itself a direct source of F1 gains beyond any single architecture.

Reading between the lines

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

  • Editorial inference: the same co-occurrence-adjusted energy mechanism could be re-expressed as a learned label-graph regularizer, removing the hand-set constant $c$ and adapting to descriptor vocabularies beyond odor.
  • Editorial inference: the paper does not report per-class F1 for rare descriptors, so whether CIL truly helps the long tail or mostly helps common descriptors remains open; a per-descriptor breakdown would settle it.
  • Editorial inference: because HMFM only modulates atomic features, bond-level and global fingerprint features are not frequency-mapped; applying the same modulation at those levels is a testable extension that may or may not add further gains.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes HMFNet, a multi-hierarchical architecture for multi-label molecular odor prediction on a combined Leffingwell/GoodScents dataset. It combines a local multi-hierarchy feature extraction module with a harmonic modulated feature mapping (HMFM), a global fingerprint/Transformer module, and a five-term Chemically-Informed Loss (CIL). The central claim, stated in Section 4.5 and supported by Table 1, is that the full model achieves state-of-the-art F1 score (0.4861) and AUROC (0.9316) compared with eight GNN baselines and four feature-mapping methods. The paper also reports ablation studies and hyperparameter sensitivity analyses in the appendix.

Significance. If the reported results hold, the paper would make a useful empirical contribution to molecular odor prediction: it combines several complementary feature sources (atomic, bond, fingerprint, SMILES-token) and demonstrates consistent F1 gains from the proposed modulation and loss across multiple backbones. The modular ablation design, with comparisons to four feature-mapping methods and five loss functions, is a strength. However, the evaluation protocol is not fully specified, and the CIL formulation contains a mathematical degeneracy that undermines the claimed mechanism. The absence of train/test split details, error bars, and significance tests means the state-of-the-art claim is not currently verifiable. The paper also ships no code or data to resolve these ambiguities.

major comments (4)
  1. [Section 4.1 and Section 3.4] The evaluation protocol is underspecified in a way that directly affects the central SOTA claim. The paper never states the train/validation/test split, the random seed, the number of runs, or whether the Leffingwell and GoodScents datasets were deduplicated before concatenation. More importantly, Eq. (10) defines E_energy(j) as the average prediction 'across the entire sample set,' and Eqs. (11)-(12) compute m_in and m_out from sums over N samples; if N refers to the full dataset rather than the training fold, the energy targets and L_class become functions of test labels, leaking label frequencies into training. The hyperparameter selections in Tables 5-6 are reported without stating which split produced the metrics, so selection on the test set cannot be excluded. The authors must specify the split, compute all label statistics and hyperparameter choices from training data only, and report mean and standard deviation over multiple seeds before the state-of-the-art claim in Section 4.5 can be accepted.
  2. [Section 3.4, Eqs. (10)-(13)] The chemical-property energy loss L_class is mathematically degenerate as written. Since m_in = 1 + c * diag((1/N) sum_i Y_i^T Y_i) and c > 0, m_in is strictly greater than 1, while E_energy(j) is an average predicted probability and hence at most 1. Therefore the first term of Eq. (13), sum over positive samples of max(0, E_energy(j) - m_in)^2, is always zero for every class. The positive-class energy target never influences training; only the second term, involving m_out, is active. This means the claimed mechanism of 'setting a target energy for each odor descriptor' is not implemented by the stated formula. The authors need to correct the definition (for example, by removing the +1 or by defining E_energy differently) and rerun the experiments, or provide a derivation showing why the first term can be nonzero.
  3. [Section 3.4, Eqs. (11)-(12)] The claim that CIL uses label co-occurrence is not supported by the formulas. Eq. (11) computes diag((1/N) sum_i Y_i^T Y_i), and Eq. (12) computes diag((1/N) sum_i (1-Y_i)^T(1-Y_i)). Taking only the diagonal discards all off-diagonal entries of the co-occurrence matrices, so m_in and m_out depend only on each descriptor's own marginal positive and negative frequencies, not on pairwise co-occurrence. The sentence in Section 3.4 stating that 'Descriptors that frequently co-occur are assigned higher energy targets' is therefore incorrect relative to the implemented equations. The authors should either use the full co-occurrence matrix (including off-diagonal terms) or revise the text to describe the loss as using per-descriptor frequencies.
  4. [Section 4.5 and Table 1] The overall claim that 'our approach achieves the best performance in molecular odor prediction' is not fully supported by the reported metrics. In Table 1, the full model achieves F1 = 0.4861 but AUROC = 0.9316, which is lower than the AUROC of 0.9356 obtained with only node and edge features. Several baselines in Table 2 also achieve higher AUROC. Since the paper explicitly defines both F1 and AUROC as evaluation metrics, the SOTA claim needs to be qualified: the method is best on F1 but not on AUROC. The authors should either report additional metrics that justify 'best performance' or adjust the claim to reflect the metric-specific nature of the result.
minor comments (4)
  1. [Figure 1 caption] The caption contains a duplicated phrase: 'The concrete structure of concrete structure of Harmonic Modulated Feature Mapping' should be 'The concrete structure of Harmonic Modulated Feature Mapping.'
  2. [Tables 5-6 and Appendix A.2] There are typographical errors in the hyperparameter tables: '0,1' should be '0.1' in Table 6, and the header 'λ1. λ2, λ3, andλ4' is missing a space before λ4. These should be corrected in a revision.
  3. [Figures 2, 5, 6] The figures are referenced but not described in the text. For example, Figure 5 and Figure 6 are only mentioned in the appendix without a discussion of how the combined dataset was constructed or how the co-occurrence matrix was thresholded. Adding a sentence or two explaining the figures would improve clarity.
  4. [Section 4.1] The paper does not report the number of molecules, the number of odor descriptors, or the exact train/test split size for the combined dataset. Providing these numbers in Section 4.1 or Appendix A.1 is essential for reproducibility and for comparing with prior work.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pipeline is an empirical supervised-learning system, and the evaluation-protocol gaps are correctness concerns rather than derivational circularity.

full rationale

The paper's derivation chain is modular and supervised. LMFE, GMFE, and HMFM are feature-extraction components trained with losses whose targets are the labels Y and whose arguments are predictions Y-hat; CIL combines weighted BCE, energy constraints, sample-level constraints, and correlation constraints, all of which are standard supervised losses (Eqs. 9, 13, 15, 16). Using label statistics to define loss targets is supervised training, not a prediction derived from those statistics by construction. The only self-citation by overlapping authors, Xie et al. 2025 for AFL, appears in Table 4 as a baseline loss comparison and is not load-bearing for HMFNet's novelty or for ruling out alternatives. The reported F1/AUROC are empirical comparisons against external baselines, so the central claim is not equivalent to its inputs by definition. Concerns about the unspecified train/test split and the fact that Eq. 10 averages over the 'entire sample set' are evaluation-protocol and reproducibility risks, not circularity; similarly, the diagonal extraction in Eqs. 11-12 being inconsistent with the stated co-occurrence mechanism is a formula-meaning issue rather than a circular derivation. No circular step is established by the quoted equations.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim depends on a dataset whose label statistics are used inside the loss, so the label distribution is an input to the model. The hyperparameters are fitted to the same data. No new physical entities are introduced.

free parameters (5)
  • c = 0.2
    Controls adjustment of energy targets m_in and m_out in CIL (Eq. 11-12); selected by maximum F1 in Table 5.
  • lambda_1..lambda_4 = 1, 0.2, 0.1, 0.2
    Weights for the four loss terms in Eq. 17; selected by maximum F1 in Table 6.
  • e1, e2 = not specified (e1+e2=1)
    Hyperparameters in the sample-level expected energy, Eq. 14; no values or tuning procedure reported.
  • sigma_prime = not specified
    Standard deviation used in the base frequency b=2*pi*sigma'*j/D in HMFM (Section 3.2); it is computed from data but the formula and source are not stated.
  • w_j (class weights) = dynamic ratio Wneg/Wpos clipped to [0.1,10]
    Per-class weights in weighted BCE, Eq. 8; recomputed from each batch, so the effective weighting varies during training.
assumptions (4)
  • domain assumption RDKit-computed atomic features, bond features, and the three fingerprint types (Morgan, MACCS, Topological) capture chemical structure relevant to odor.
    Invoked in Sections 3.2 and 3.3 without justification; the entire input representation rests on this.
  • domain assumption The co-occurrence statistics of odor descriptors computed from the dataset reflect stable label dependencies that improve generalization.
    Used to build m_in and m_out in Eq. 11-12 and the label correlation loss Eq. 16; no evidence that these statistics are stable across splits.
  • domain assumption The combined Leffingwell and GoodScents dataset is representative of molecular odor space and the labels are reliable.
    All experiments use this dataset; no external validation is performed.
  • domain assumption A Transformer over SMILES strings and a GAT over the molecular graph can learn complementary representations.
    Assumed in the architecture design, Section 3.3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Hierarchical Fine-Grained Feature Mapping Driven by Feature Contribution for Molecular Odor Prediction." pith.science (2026). https://pith.science/paper/M62U253B

@misc{pith2026250500290,
  author       = {Pith},
  title        = {Pith review of: Multi-Hierarchical Fine-Grained Feature Mapping Driven by Feature Contribution for Molecular Odor Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M62U253B}},
  note         = {Machine review of arXiv:2505.00290}
}
read the original abstract

Molecular odor prediction is the process of using a molecule's structure to predict its smell. While accurate prediction remains challenging, AI models can suggest potential odors. Existing methods, however, often rely on basic descriptors or handcrafted fingerprints, which lack expressive power and hinder effective learning. Furthermore, these methods suffer from severe class imbalance, limiting the training effectiveness of AI models. To address these challenges, we propose a Feature Contribution-driven Hierarchical Multi-Feature Mapping Network (HMFNet). Specifically, we introduce a fine-grained, Local Multi-Hierarchy Feature Extraction module (LMFE) that performs deep feature extraction at the atomic level, capturing detailed features crucial for odor prediction. To enhance the extraction of discriminative atomic features, we integrate a Harmonic Modulated Feature Mapping (HMFM). This module dynamically learns feature importance and frequency modulation, improving the model's capability to capture relevant patterns. Additionally, a Global Multi-Hierarchy Feature Extraction module (GMFE) is designed to learn global features from the molecular graph topology, enabling the model to fully leverage global information and enhance its discriminative power for odor prediction. To further mitigate the issue of class imbalance, we propose a Chemically-Informed Loss (CIL). Experimental results demonstrate that our approach significantly improves performance across various deep learning models, highlighting its potential to advance molecular structure representation and accelerate the development of AI-driven technologies.

Figures

Figures reproduced from arXiv: 2505.00290 by the authors.

Figure 1
Figure 1. (a): The concrete structure of fine-grained Local Multi-Hierarchical Feature Extraction; (b): The concrete structure of Global [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Co-ocurrence matrix of Top 20 odor descriptors. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Comparison of F1 scores of histogram of Harmonic Mod [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: The combined dataset benefits from the strengths [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 5
Figure 5. Figure 5: Distribution of odor descriptor frequency in dataset. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Co-ocurrence matrix for odor descriptors. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 27 canonical work pages

  1. [1]

    Incorporating label dependency into the binary relevance framework for multi-label classification

    [Alvares-Cherman et al., 2012] Everton Alvares-Cherman, Jean Metz, and Maria Carolina Monard. Incorporating label dependency into the binary relevance framework for multi-label classification. Expert Systems with Applications, 39(2):1647–1655,

    work page 2012

  2. [9]

    Predicting human olfactory percep- tion from chemical features of odor molecules

    [Keller et al., 2017] Andreas Keller, Richard C Gerkin, Yuanfang Guan, Amit Dhurandhar, Gabor Turu, Bence Szalai, Joel D Mainland, Yusuke Ihara, Chung Wen Yu, Russ Wolfinger, et al. Predicting human olfactory percep- tion from chemical features of odor molecules. Science, 355(6327):820–826,

    work page 2017

  3. [10]

    Hierarchy-aware biased bound margin loss func- tion for hierarchical text classification

    [Kim et al., 2024] Gibaeg Kim, SangHun Im, and Heung- Seon Oh. Hierarchy-aware biased bound margin loss func- tion for hierarchical text classification. In Findings of the Association for Computational Linguistics ACL 2024 , pages 7672–7682,

    work page 2024

  4. [11]

    Semi-supervised classification with graph convolutional networks

    [Kipf and Welling, 2016] Thomas N Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907,

    arXiv 2016

  5. [13]

    Unsupervised Optimisation of GNNs for Node Clustering

    [Leeney and McConville, 2024] William Leeney and Ryan McConville. An investigation into using unsupervised metrics to optimise gnns for node clustering. arXiv preprint arXiv:2402.07845,

    work page Pith review arXiv 2024

  6. [14]

    Leffingwell & as- sociates

    [Leffingwell, 2005] John C Leffingwell. Leffingwell & as- sociates. Chirality & Odour Perception. Available on- line: http://www. leffingwell. com/chirality/chirality. htm (accessed on 15 November 2023),

    work page 2005

  7. [16]

    Random fourier feature-based deep learning for wireless communications

    [Mitra and Kaddoum, 2022] Rangeet Mitra and Georges Kaddoum. Random fourier feature-based deep learning for wireless communications. IEEE Transactions on Cog- nitive Communications and Networking , 8(2):468–479,

    work page 2022

  8. [17]

    Mordred: a molecular descriptor calculator

    [Moriwaki et al., 2018] Hirotomo Moriwaki, Yu-Shi Tian, Norihito Kawashita, and Tatsuya Takagi. Mordred: a molecular descriptor calculator. Journal of cheminformat- ics, 10:1–14,

    work page 2018

Show all 36 references
  1. [18]

    Topo- logical torsion: a new molecular descriptor for sar ap- plications

    [Nilakantan et al., 1987] Ramaswamy Nilakantan, Norman Bauman, J Scott Dixon, and R Venkataraghavan. Topo- logical torsion: a new molecular descriptor for sar ap- plications. comparison with other descriptors. Journal of Chemical Information and Computer Sciences , 27(2):82– 85,

    1987

  2. [20]

    Extended-connectivity fingerprints

    [Rogers and Hahn, 2010] David Rogers and Mathew Hahn. Extended-connectivity fingerprints. Journal of chemical information and modeling, 50(5):742–754,

    2010

  3. [22]

    Owsum: algo- rithmic odor prediction and insight into structure-odor re- lationships

    [Schicker et al., 2023] Doris Schicker, Satnam Singh, Jes- sica Freiherr, and Andreas T Grasskamp. Owsum: algo- rithmic odor prediction and insight into structure-odor re- lationships. Journal of Cheminformatics, 15(1):51,

    2023

  4. [23]

    Smiles to smell: decoding the structure–odor relationship of chemical com- pounds using the deep neural network approach

    [Sharma et al., 2021] Anju Sharma, Rajnish Kumar, Shab- nam Ranjta, and Pritish Kumar Varadwaj. Smiles to smell: decoding the structure–odor relationship of chemical com- pounds using the deep neural network approach. Journal of Chemical Information and Modeling , 61(2):676–688,

    2021

  5. [24]

    Attention is all you need

    [Vaswani, 2017] A Vaswani. Attention is all you need. Ad- vances in Neural Information Processing Systems,

    2017

  6. [25]

    Graph attention networks

    [Veliˇckovi´c et al., 2017] Petar Veliˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. Graph attention networks. arXiv preprint arXiv:1710.10903,

    2017 arXiv

  7. [26]

    Local random feature approximations of the gaussian kernel

    [Wacker and Filippone, 2022] Jonas Wacker and Maurizio Filippone. Local random feature approximations of the gaussian kernel. Procedia Computer Science , 207:987– 996,

    2022

  8. [27]

    Distributionally robust beamforming and estimation of wireless signals

    [Wang et al., 2024b] Shixiong Wang, Wei Dai, and Ge- offrey Ye Li. Distributionally robust beamforming and estimation of wireless signals. arXiv preprint arXiv:2401.12345,

    arXiv

  9. [28]

    An explainable deep learning platform for molecular discovery

    [Wong et al., 2024] Felix Wong, Satotaka Omori, Alicia Li, Aarti Krishnan, Ryan S Lach, Joseph Rufo, Maxwell Z Wilson, and James J Collins. An explainable deep learning platform for molecular discovery. Nature Protocols, pages 1–37,

    2024

  10. [29]

    A com- prehensive survey on graph neural networks

    [Wu et al., 2020] Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. A com- prehensive survey on graph neural networks. IEEE trans- actions on neural networks and learning systems, 32(1):4– 24,

    2020

  11. [33]

    Padel-descriptor: An open source software to calculate molecular descriptors and fingerprints

    [Yap, 2011] Chun Wei Yap. Padel-descriptor: An open source software to calculate molecular descriptors and fingerprints. Journal of computational chemistry , 32(7):1466–1474,

    2011

  12. [35]

    A machine learning based computer-aided molecular design/screening methodology for fragrance molecules

    [Zhang et al., 2018] Lei Zhang, Haitao Mao, Linlin Liu, Jian Du, and Rafiqul Gani. A machine learning based computer-aided molecular design/screening methodology for fragrance molecules. Computers & Chemical Engi- neering, 115:295–308,

    2018

  13. [1987]

    Asymmetric loss for multi-label clas- sification

    [Ridnik et al., 2021] Tal Ridnik, Emanuel Ben-Baruch, Na- dav Zamir, Asaf Noy, Itamar Friedman, Matan Protter, and Lihi Zelnik-Manor. Asymmetric loss for multi-label clas- sification. In Proceedings of the IEEE/CVF international conference on computer vision, pages 82–91,

    2021

  14. [2001]

    fruity”, “floral

    The Leffingwell PMP 2001 is a widely used resource in the study of molecular odor descrip- tors. It consists of a collection of chemical compounds along with their corresponding odor descriptors, which are used to capture the sensory characteristics of various molecules. The d...

    2001

  15. [2002]

    Food, and cosmetics ingre- dients information

    [Flavor, 2018] Fragrance Flavor. Food, and cosmetics ingre- dients information. The Good Scents Company,

    2018

  16. [2010]

    Predicting odor from molecular structure: A multi-label classification approach

    [Saini and Ramanathan, 2022] Kushagra Saini and Venkat- narayan Ramanathan. Predicting odor from molecular structure: A multi-label classification approach. Scientific reports, 12(1):13863,

    2022

  17. [2011]

    Prediction of anticancer peptides based on an ensemble model of deep learning and machine learning using or- dinal positional encoding

    [Yuan et al., 2023] Qitong Yuan, Keyi Chen, Yimin Yu, Nguyen Quoc Khanh Le, and Matthew Chin Heng Chua. Prediction of anticancer peptides based on an ensemble model of deep learning and machine learning using or- dinal positional encoding. Briefings in Bioinformatics , 24(1):bbac630,

    2023

  18. [2012]

    Balanced energy regularization loss for out-of-distribution detection

    [Choi et al., 2023] Hyunjun Choi, Hawook Jeong, and Jin Young Choi. Balanced energy regularization loss for out-of-distribution detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15691–15700,

    2023

  19. [2016]

    A principal odor map unifies diverse tasks in olfactory perception

    [Lee et al., 2023] Brian K Lee, Emily J Mayhew, Benjamin Sanchez-Lengeling, Jennifer N Wei, Wesley W Qian, Kelsie A Little, Matthew Andres, Britney B Nguyen, Theresa Moloy, Jacob Yasonik, et al. A principal odor map unifies diverse tasks in olfactory perception. Science, 381(6...

    2023

  20. [2017]

    Few-shot graph learning for molecular property predic- tion

    [Guo et al., 2021] Zhichun Guo, Chuxu Zhang, Wenhao Yu, John Herr, Olaf Wiest, Meng Jiang, and Nitesh V Chawla. Few-shot graph learning for molecular property predic- tion. In Proceedings of the web conference 2021 , pages 2559–2567,

    2021

  21. [2018]

    Neu- ral message passing for quantum chemistry

    [Gilmer et al., 2017] Justin Gilmer, Samuel S Schoenholz, Patrick F Riley, Oriol Vinyals, and George E Dahl. Neu- ral message passing for quantum chemistry. In Interna- tional conference on machine learning, pages 1263–1272. PMLR,

    2017

  22. [2019]

    Spatial smoothing using graph laplacian penalized filter

    [Yamada, 2024] Hiroshi Yamada. Spatial smoothing using graph laplacian penalized filter. Spatial Statistics , page 100799,

    2024

  23. [2020]

    Molecu- lar odor prediction based on multi-feature graph attention networks

    [Xie et al., 2025] HongXin Xie, JianDe Sun, Yi Shao, Shuai Li, Sujuan Hou, YuLong Sun, and Jian Wang. Molecu- lar odor prediction based on multi-feature graph attention networks. arXiv preprint arXiv:2502.01430,

    2025 arXiv

  24. [2021]

    Sub- graph generation applied in graphsage deal with imbal- anced node classification

    [Huang and Chen, 2024] Kai Huang and Chen Chen. Sub- graph generation applied in graphsage deal with imbal- anced node classification. Soft Computing , pages 1–14,

    2024

  25. [2022]

    Cross-entropy loss functions: Theoretical analysis and applications

    [Mao et al., 2023] Anqi Mao, Mehryar Mohri, and Yutao Zhong. Cross-entropy loss functions: Theoretical analysis and applications. In International conference on Machine learning, pages 23803–23828. PMLR,

    2023

  26. [2023]

    Reoptimization of mdl keys for use in drug discovery

    [Durant et al., 2002] Joseph L Durant, Burton A Leland, Douglas R Henry, and James G Nourse. Reoptimization of mdl keys for use in drug discovery. Journal of chemi- cal information and computer sciences, 42(6):1273–1280,

    2002

  27. [2024]

    Recent ad- vances and application of machine learning in food flavor prediction and regulation

    [Ji et al., 2023] Huizhuo Ji, Dandan Pu, Wenjing Yan, Qingchuan Zhang, Min Zuo, and Yuyu Zhang. Recent ad- vances and application of machine learning in food flavor prediction and regulation. Trends in Food Science & Tech- nology, 138:738–751,

    2023

  28. [2025]

    Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism

    [Xiong et al., 2019] Zhaoping Xiong, Dingyan Wang, Xi- aohong Liu, Feisheng Zhong, Xiaozhe Wan, Xutong Li, Zhaojun Li, Xiaomin Luo, Kaixian Chen, Hualiang Jiang, et al. Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism. Jo...

    2019

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.