Pith. sign in

REVIEW 4 major objections 7 minor 33 references

An accuracy improving method for advertising click through rate prediction based on enhanced xDeepFM model

T0 review · 4 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that adding a multi-head attention layer and replacing xDeepFM's linear component with a Factorization Machine improves CTR prediction accuracy on the Criteo dataset.

desk verdict The architecture is a reasonable incremental variant, but the test-set tuning makes the reported gains untrustworthy and the abstract overstates the results. read the letter →

arxiv 2411.15223 v1 pith:NRA2F63T submitted 2024-11-21 cs.LG

Xiaowei Xi , Song Leng , Yuqing Gong , Dalin Li This is my paper · ORCID
classification cs.LG
keywords CTRpredictionxDeepFMmulti-headattentionfactorizationmachineCriteodatasetAUCLoglossfeatureinteraction
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 an enhanced xDeepFM model for advertising click-through rate prediction and reports that it outperforms the original xDeepFM, DeepFM, and several other baselines on a 500,000-sample Criteo subset. The two modifications are a multi-head attention layer placed before the deep network, with residual connections and layer normalization, and a Factorization Machine replacing the linear component. On the reported test set the model reaches AUC 0.7850 and Logloss 0.4628, a gain of 0.0072 AUC and 0.0061 Logloss over xDeepFM. If the result holds, it offers a relatively contained way to improve CTR prediction accuracy on sparse, high-dimensional advertising data.

What carries the argument

The two load-bearing additions are the multi-head attention layer, computed as $\text{MultiHead}(Q,K,V)=\text{Concat}(\text{head}_1,\ldots,\text{head}_h)W^O$ with each head an attention operation, and the Factorization Machine replacing the linear term, with prediction $\hat{y}(x)=w_0+\sum_i w_i x_i+\sum_i\sum_{j>i}\langle v_i,v_j\rangle x_i x_j$. The attention layer is inserted ahead of the DNN with residual connections and layer normalization so original embedding information is retained while feature correlations are captured from multiple subspaces. The FM branch contributes low-order interaction modeling that the linear branch of xDeepFM cannot express. These outputs are combined with the CIN and DNN outputs and mapped by a sigmoid to produce the final click probability.

What would settle it

Retrain the proposed model and xDeepFM on Criteo with a train/validation/test split, tune only on validation, and report the best-validation-epoch test AUC and Logloss across several random seeds. If the AUC advantage over xDeepFM does not reproduce on a test set that was never used for model selection, the central claim of significant improvement is not supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that its enhanced xDeepFM architecture predicts ad clicks more accurately than the model it builds on. The final prediction is the sigmoid-weighted fusion of three branches: an FM branch for first- and second-order feature interactions, a Compressed Interaction Network for explicit high-order interactions, and a DNN branch whose input embeddings are first passed through a multi-head attention layer with residuals and layer normalization. On the Criteo benchmark the paper reports test AUC 0.7850 and Logloss 0.4628, compared with 0.7778 and 0.4689 for xDeepFM, and interprets these differences as evidence that both modifications contribute to better modeling of feature relationships in sparse ad data.

Load-bearing premise

The load-bearing premise is that the hyperparameters chosen by inspecting AUC and Logloss on the test set in Sections 3.3.1 through 3.3.3 generalize to the final reported test numbers; if they do not, the 0.0072 AUC gain could reflect selection bias rather than a genuine improvement in the model.

Editorial extensions

If this is right

  • If the reported results are accepted, porting these two modifications into an existing xDeepFM-based CTR system yields a gain of roughly 0.0072 AUC on a large, sparse advertising dataset.
  • The lower Logloss alongside the higher AUC indicates that predicted click probabilities become both better ranked and better calibrated.
  • Because the attention layer is preposed and the DNN and CIN components are unchanged, the modification can be layered onto similar feature-interaction architectures without redesigning their training pipelines.
  • The combination of FM, CIN, and attention-enhanced DNN lets one network cover first-order, second-order, explicit high-order, and implicit high-order feature interactions simultaneously.

Reading between the lines

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

  • A direct testable extension is to repeat the hyperparameter search using a separate validation split and touch the test set only once; the reported gain would be materially stronger evidence if it survives that protocol.
  • The baselines SEDAFM, MAME-DFM, DTM, and DMCNN are compared using numbers from other sources rather than retrained under identical conditions, so an apples-to-apples retraining comparison would clarify whether the improvement is specific to this implementation.
  • The claim that computational complexity does not significantly increase is stated, not measured; a parameter-count and runtime comparison against xDeepFM would make it testable.
  • Re-running the same two modifications on other sparse CTR datasets or larger Criteo samples, with multiple seeds, would show whether the 0.0072 AUC gain is stable or dataset-specific.
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.

Referee Report

4 major / 7 minor

Summary. The paper proposes an enhanced xDeepFM model for advertising click-through rate prediction, replacing the linear component with a Factorization Machine and adding a multi-head attention layer. The authors evaluate the model on a 500,000-sample subset of the Criteo dataset, report AUC and Logloss, and claim that their model outperforms DeepFM, xDeepFM, and other state-of-the-art methods. The core claim is unsupported because the final test metrics are obtained from the same test set used to select hyperparameters, and Table 3 internally contradicts the claim of universal superiority.

Significance. If the claimed improvements were established with a sound evaluation protocol, the architectural modifications (multi-head attention plus FM-based low-order interaction modeling) would be a modest but potentially useful incremental contribution to CTR prediction research. However, the paper provides no credible evidence for this claim: the evaluation is methodologically invalid, the results are not reproducible from the information given, and the reported numbers are not internally consistent. The paper ships no code, no detailed preprocessing steps, and no statistical significance analysis, so even the descriptive contribution is weakened. The central idea is not original enough to carry the paper by itself without a trustworthy empirical demonstration.

major comments (4)
  1. [§3.3.1–§3.3.4] The experimental protocol invalidates the headline claim. Sections 3.3.1 through 3.3.3 select the learning rate (0.05), embedding size (8), and number of attention heads (2) by comparing AUC and Logloss on the test set, and Section 3.3.4 then reports the final comparison on that same test set. Because the test set has already been used for model selection, the final metrics are not independent; best-of-test selection can create a spurious advantage over baselines that were not tuned on the same split. The paper must either use a separate validation set for hyperparameter selection and reserve the test set for a single final evaluation, or use nested cross-validation. Without this, the reported 0.0072 AUC improvement over xDeepFM cannot be attributed to the architecture.
  2. [Table 3 and Abstract] The broad claim that the model outperforms other state-of-the-art methods is contradicted by the paper's own Table 3. The proposed model achieves AUC 0.7850 and Logloss 0.4628, but DMCNN achieves a higher AUC (0.7994) and DTM achieves a lower Logloss (0.4488) on the same reported benchmark. The abstract and Section 3.3.4 state that the model outperforms other state-of-the-art methods without qualification, which is not supported by these data. The authors should restrict their claims to the two baselines actually compared in their own experiments, or explain why the other rows are not considered in the comparison.
  3. [§3.3.4] No measures of uncertainty are reported. All comparisons are single point estimates with no error bars, no multiple seeds, and no statistical significance tests. Given that the reported difference in AUC between the proposed model and xDeepFM is only 0.0072, and the difference between runs with different hyperparameter settings in Tables 1 and 2 is of a similar magnitude (e.g., 0.7849 vs. 0.7850), the claimed improvement may be within run-to-run noise. The authors should report mean and standard deviation over multiple random seeds and perform a paired significance test (e.g., a paired bootstrap or Wilcoxon test) on the test predictions.
  4. [§3.1 and §3.3.4] The experimental description is missing crucial reproducibility details. The paper states that 500,000 samples were used for training, but it is unclear whether the full sampled dataset is 500,000 samples split 80/20 or whether 500,000 is the training size only. There is no description of how missing values were handled, how the categorical features were one-hot encoded or embedded, what the feature cardinalities are, or how the baseline models (DeepFM, xDeepFM, SEDAFM, MAME-DFM, DTM, DMCNN) were configured and tuned. Without these details, no one can reproduce or verify the results, and the comparison may not be fair if the baselines were not tuned with the same effort.
minor comments (7)
  1. [Table 2] The column header 'Number of Long Attention Head' appears to be a typo; it should read 'Number of Attention Heads'.
  2. [§3.3.4] The sentence 'The test results of different models on the same dataset are shown in Table 3 and Figures 2 and 3' is confusing because Figures 2 and 3 are the architecture diagrams of CIN and the multi-head attention mechanism, not experimental result plots. The reference should be corrected to point to the relevant result figures or be removed.
  3. [§3.2] The AUC formula in Eq. (8) is written as a single line without the rank-sum variable on the left-hand side; the notation should be cleaned up and the 'rank' variable explained consistently.
  4. [§3.3.1–§3.3.3] The hyperparameter experiments are presented as if the best configuration is unambiguous, but the differences are very small (e.g., AUC 0.7849 at embedding size 8 versus 0.7844 at size 6). The authors should acknowledge that these differences may not be practically significant and would benefit from statistical testing even for the hyperparameter choice.
  5. [Abstract and §3.3.4] The abstract claims that the improvements are achieved 'without significantly increasing computational complexity,' but no runtime, parameter count, or computational cost experiments are reported. This claim should either be substantiated or removed.
  6. [References] Several references are incomplete or inconsistently formatted; for example, Vaswani (2017), Hochreiter (1997), and O'Shea (2015) lack full author lists and titles, while other entries include them. The reference list should be brought to a consistent, complete format.
  7. [Throughout] The manuscript contains numerous typographical and formatting issues ('School of Dig Data', 'X DEEP FM', garbled equations, and inconsistent use of styles in the formulas). A thorough proofreading and LaTeX cleanup is needed before publication.

Circularity Check

1 steps flagged · score 5.0 of 10

Final AUC/Logloss are selected on the same test split, so the reported architecture gain is not an independent prediction.

  1. fitted input called prediction [Section 3.3, Sections 3.3.1-3.3.4]
    "In the experiments, the dataset was divided into training and test sets in an 8:2 ratio. ... According to the AUC and Logloss results in Figure 5, the model achieved the best prediction performance when the learning rate was 0.05. ... According to Table 1, when the embedding vector dimension was 8, the AUC value was 0.7849, and the Logloss value was 0.4629, indicating optimal prediction performance. ... when the number of attention heads was 2, the AUC value was 0.7850, and the Logloss value was 0.4628, indicating optimal prediction performance."

    Sections 3.3.1-3.3.3 tune learning rate, embedding size, and attention-head count by comparing AUC/Logloss on the same 20% test split described at the start of Section 3.3. Section 3.3.4 then reports exactly those best-test values (AUC 0.7850, Logloss 0.4628) as the final model performance and attributes the 0.0072 AUC gain over xDeepFM to the architecture. Because the hyperparameters were chosen to maximize performance on this exact split, the reported final metrics are selection results rather than independent predictions. The claimed improvement is therefore partly forced by the tuning procedure itself, not derived solely from the model design.

full rationale

The derivation chain of the model design itself (multi-head attention plus FM replacing the linear component) is not circular: the architecture is a genuine modification of xDeepFM, and the paper does not rely on a self-cited uniqueness theorem or redefine a known result. The only substantial circularity-type defect is in the evaluation protocol. The same 8:2 training/test split is used both for hyperparameter selection (learning rate, embedding dimension, number of heads) and for the final reported comparison, so the best-of-test configuration is not a held-out prediction. With multiple configurations evaluated on the same test set, the reported 0.7850 AUC and 0.4628 Logloss can reflect selection bias, and the 0.0072 AUC improvement over the untuned xDeepFM baseline is not an unbiased estimate. The paper also reports no separate validation set, no multiple seeds, and no error bars. Additionally, the abstract's claim that the model 'outperforms other state-of-the-art methods' is contradicted by the paper's own Table 3, where DMCNN has higher AUC (0.7994) and DTM has lower Logloss (0.4488) than the proposed model; this is a consistency issue rather than circularity. On balance, the central architecture claim has independent content, but the central reported prediction is compromised by test-set tuning, giving a moderate circularity score.

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

The central claim rests on test-set hyperparameter selection, an unverified 500k-sample assumption, and unvalidated baseline numbers; there is no code, no external benchmark, and no formal or statistical evidence.

free parameters (5)
  • learning_rate = 0.05
    Chosen by grid search on the test set in Section 3.3.1, not justified by theory or validation.
  • embedding_dimension = 8
    Selected by test-set comparison in Section 3.3.2; only 2, 4, 6, and 8 were tried.
  • attention_heads = 2
    Selected by test-set comparison in Section 3.3.3; only 1, 2, 4, and 8 were tried.
  • dnn_hidden_units = (256, 128)
    Set in Section 3.3.4 with no ablation and no sensitivity analysis.
  • batch_size = 2048 train, 4096 test
    Chosen in Section 3.3.4 with no stated justification.
assumptions (4)
  • domain assumption The 500k stratified sample of Criteo preserves the full-data ranking difficulty and transfers to real ad traffic.
    Section 3.1 states stratified sampling was used but gives no evidence that conclusions generalize beyond this subsample.
  • ad hoc to paper Hyperparameters selected on the test set remain optimal for unseen data.
    Sections 3.3.1 through 3.3.3 tune hyperparameters on the same 20% test set later used for final metrics, violating standard train, validation, test separation.
  • domain assumption Multi-head attention and FM improve xDeepFM without harming CIN feature interactions.
    The paper asserts this in Sections 2.1 and 2.2 but provides no ablation isolating the marginal contribution of each component.
  • standard math AUC and Logloss as defined in Equations 8 and 9 are valid for the binary classification setting.
    These are standard definitions; no issue with the formulas themselves.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An accuracy improving method for advertising click through rate prediction based on enhanced xDeepFM model." pith.science (2026). https://pith.science/paper/NRA2F63T

@misc{pith2026241115223,
  author       = {Pith},
  title        = {Pith review of: An accuracy improving method for advertising click through rate prediction based on enhanced xDeepFM model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NRA2F63T}},
  note         = {Machine review of arXiv:2411.15223}
}
read the original abstract

Advertising click-through rate (CTR) prediction aims to forecast the probability that a user will click on an advertisement in a given context, thus providing enterprises with decision support for product ranking and ad placement. However, CTR prediction faces challenges such as data sparsity and class imbalance, which adversely affect model training effectiveness. Moreover, most current CTR prediction models fail to fully explore the associations among user history, interests, and target advertisements from multiple perspectives, neglecting important information at different levels. To address these issues, this paper proposes an improved CTR prediction model based on the xDeepFM architecture. By integrating a multi-head attention mechanism, the model can simultaneously focus on different aspects of feature interactions, enhancing its ability to learn intricate patterns without significantly increasing computational complexity. Furthermore, replacing the linear model with a Factorization Machine (FM) model improves the handling of high-dimensional sparse data by flexibly capturing both first-order and second-order feature interactions. Experimental results on the Criteo dataset demonstrate that the proposed model outperforms other state-of-the-art methods, showing significant improvements in both AUC and Logloss metrics. This enhancement facilitates better mining of implicit relationships between features and improves the accuracy of advertising CTR prediction.

Figures

Figures reproduced from arXiv: 2411.15223 by the authors.

Figure 1
Figure 1. Overall architecture of the improved xDeepFM model incorporating the multi-head attention mechanism and [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The Architecture of CIN 2.3.1 Multi-head Attention Layer Inspired by the research of Song et al. Song et al. [2019], we combine the attention mechanism with the xDeepFM model, introducing a preposed multi-head attention layer into its architecture, as shown in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Structural diagram of multi-head attention mechanism [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Distribution of data feature types 3.2 Evaluation Metrics CTR prediction is essentially a binary classification problem Deng et al. [2021]. To evaluate the model’s effectiveness, this paper uses AUC and Logloss as evaluation metrics. AUC measures the ability of the mod…
Figure 5
Figure 5. Figure 5: Summary of the impact of different learning rates on the predicted outcomes [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Effect of different Embedding vector dimensions on the prediction results [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Effect of different attention head counts on predicted outcomes [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 25 canonical work pages

  1. [1]

    Cracking under pressure: current trends in the global advertising industry

    John Sinclair. Cracking under pressure: current trends in the global advertising industry. Media International Australia, 174 0 (1): 0 3--16, 2020

    work page 2020

  2. [2]

    Power of digital marketing in building brands: A review of social media advertisement

    Rammohan Sundaram, Dr Rajeev Sharma, and Dr Anurag Shakya. Power of digital marketing in building brands: A review of social media advertisement. International Journal of Management, 11 0 (4), 2020

    work page 2020

  3. [3]

    Improving internet advertising using click--through rate prediction

    Rakesh Gudipudi, Sandra Nguyen, Doina Bein, and Sudarshan Kurwadkar. Improving internet advertising using click--through rate prediction. Applied Human Factors and Ergonomics International, 2023

    work page 2023

  4. [4]

    Consumer understanding and advertising strategy: analysis and strategic translation of laddering data

    Charles E Gengler and Thomas J Reynolds. Consumer understanding and advertising strategy: analysis and strategic translation of laddering data. Journal of advertising research, 35 0 (4): 0 19--34, 1995

    work page 1995

  5. [5]

    The online advertising industry: Economics, evolution, and privacy

    David S Evans. The online advertising industry: Economics, evolution, and privacy. Journal of economic perspectives, 23 0 (3): 0 37--60, 2009

    work page 2009

  6. [6]

    A review of methods utilized in measuring affect, feelings, and emotion in advertising

    Judith A Wiles and T Bettina Cornwell. A review of methods utilized in measuring affect, feelings, and emotion in advertising. Current Issues and Research in Advertising, 13 0 (1-2): 0 241--275, 1991

    work page 1991

  7. [7]

    Maximizing online marketing strategy with digital advertising

    Dewi Surya Wuisan and Tessa Handra. Maximizing online marketing strategy with digital advertising. Startupreneur Business Digital (SABDA Journal), 2 0 (1): 0 22--30, 2023

    work page 2023

  8. [8]

    Platform strategies and user stickiness in the online video industry

    Ke Rong, Fei Xiao, Xiaoyu Zhang, and Jingjing Wang. Platform strategies and user stickiness in the online video industry. Technological Forecasting and Social Change, 143: 0 249--259, 2019

    work page 2019

Show all 33 references
  1. [9]

    Applied logistic regression

    David W Hosmer Jr, Stanley Lemeshow, and Rodney X Sturdivant. Applied logistic regression. John Wiley & Sons, 2013

    2013

  2. [10]

    Multi-layered gradient boosting decision trees

    Ji Feng, Yang Yu, and Zhi-Hua Zhou. Multi-layered gradient boosting decision trees. Advances in neural information processing systems, 31, 2018

    2018

  3. [11]

    Factorization machines

    Steffen Rendle. Factorization machines. In 2010 IEEE International conference on data mining, pages 995--1000. IEEE, 2010

    2010

  4. [12]

    Practical lessons from predicting clicks on ads at facebook

    Xinran He, Junfeng Pan, Ou Jin, Tianbing Xu, Bo Liu, Tao Xu, Yanxin Shi, Antoine Atallah, Ralf Herbrich, Stuart Bowers, et al. Practical lessons from predicting clicks on ads at facebook. In Proceedings of the eighth international workshop on data mining for online advertising...

    2014

  5. [13]

    Learning piece-wise linear models from large scale data for ad click prediction

    Kun Gai, Xiaoqiang Zhu, Han Li, Kai Liu, and Zhe Wang. Learning piece-wise linear models from large scale data for ad click prediction. arXiv preprint arXiv:1704.05194, 2017

    2017 arXiv

  6. [14]

    Computer vision and natural language processing: recent approaches in multimedia and robotics

    Peratham Wiriyathammabhum, Douglas Summers-Stay, Cornelia Ferm \"u ller, and Yiannis Aloimonos. Computer vision and natural language processing: recent approaches in multimedia and robotics. ACM Computing Surveys (CSUR), 49 0 (4): 0 1--44, 2016

    2016

  7. [15]

    Machine learning algorithms-a review

    Batta Mahesh. Machine learning algorithms-a review. International Journal of Science and Research (IJSR).[Internet], 9 0 (1): 0 381--386, 2020

    2020

  8. [16]

    Long short-term memory

    S Hochreiter. Long short-term memory. Neural Computation MIT-Press, 1997

    1997

  9. [17]

    An introduction to convolutional neural networks

    K O'Shea. An introduction to convolutional neural networks. arXiv preprint arXiv:1511.08458, 2015

    2015 arXiv

  10. [18]

    Feed-forward neural networks

    George Bebis and Michael Georgiopoulos. Feed-forward neural networks. Ieee Potentials, 13 0 (4): 0 27--31, 1994

    1994

  11. [19]

    Wide & deep learning for recommender systems

    Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, et al. Wide & deep learning for recommender systems. In Proceedings of the 1st workshop on deep learning for recommender systems, pag...

    2016

  12. [20]

    Deepfm: a factorization-machine based neural network for ctr prediction

    Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. Deepfm: a factorization-machine based neural network for ctr prediction. arXiv preprint arXiv:1703.04247, 2017

    2017 arXiv

  13. [21]

    xdeepfm: Combining explicit and implicit feature interactions for recommender systems

    Jianxun Lian, Xiaohuan Zhou, Fuzheng Zhang, Zhongxia Chen, Xing Xie, and Guangzhong Sun. xdeepfm: Combining explicit and implicit feature interactions for recommender systems. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining, ...

    2018

  14. [22]

    Deep & cross network for ad click predictions

    Ruoxi Wang, Bin Fu, Gang Fu, and Mingliang Wang. Deep & cross network for ad click predictions. In Proceedings of the ADKDD'17 , pages 1--7. 2017

    2017

  15. [23]

    A survey on csi-based human behavior recognition in through-the-wall scenario

    Zhengjie Wang, Kangkang Jiang, Yushan Hou, Zehua Huang, Wenwen Dou, Chengming Zhang, and Yinjing Guo. A survey on csi-based human behavior recognition in through-the-wall scenario. IEEE Access, 7: 0 78772--78793, 2019

    2019

  16. [24]

    Efficient partition of integer optimization problems with one-hot encoding

    Shuntaro Okada, Masayuki Ohzeki, and Shinichiro Taguchi. Efficient partition of integer optimization problems with one-hot encoding. Scientific reports, 9 0 (1): 0 13036, 2019

    2019

  17. [25]

    Deep learning for click-through rate estimation

    Weinan Zhang, Jiarui Qin, Wei Guo, Ruiming Tang, and Xiuqiang He. Deep learning for click-through rate estimation. arXiv preprint arXiv:2104.10584, 2021

    2021 arXiv

  18. [26]

    Autoint: Automatic feature interaction learning via self-attentive neural networks

    Weiping Song, Chence Shi, Zhiping Xiao, Zhijian Duan, Yewen Xu, Ming Zhang, and Jian Tang. Autoint: Automatic feature interaction learning via self-attentive neural networks. In Proceedings of the 28th ACM international conference on information and knowledge management, pages...

    2019

  19. [27]

    Attention is all you need

    A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017

    2017

  20. [28]

    A spatial--temporal depth-wise residual network for crop sub-pixel mapping from modis images

    Yuxian Wang, Yuan Fang, Wenlong Zhong, Rongming Zhuo, Junhuan Peng, and Linlin Xu. A spatial--temporal depth-wise residual network for crop sub-pixel mapping from modis images. Remote Sensing, 14 0 (21): 0 5605, 2022

    2022

  21. [29]

    Deeplight: Deep lightweight feature interactions for accelerating ctr predictions in ad serving

    Wei Deng, Junwei Pan, Tian Zhou, Deguang Kong, Aaron Flores, and Guang Lin. Deeplight: Deep lightweight feature interactions for accelerating ctr predictions in ad serving. In Proceedings of the 14th ACM international conference on Web search and data mining, pages 922--930, 2021

    2021

  22. [30]

    Auc: a misleading measure of the performance of predictive distribution models

    Jorge M Lobo, Alberto Jim \'e nez-Valverde, and Raimundo Real. Auc: a misleading measure of the performance of predictive distribution models. Global ecology and Biogeography, 17 0 (2): 0 145--151, 2008

    2008

  23. [31]

    The use of the area under the roc curve in the evaluation of machine learning algorithms

    Andrew P Bradley. The use of the area under the roc curve in the evaluation of machine learning algorithms. Pattern recognition, 30 0 (7): 0 1145--1159, 1997

    1997

  24. [32]

    Neural networks for pattern recognition

    Christopher M Bishop. Neural networks for pattern recognition. Oxford university press, 1995

    1995

  25. [33]

    Deep metric learning via lifted structured feature embedding

    Hyun Oh Song, Yu Xiang, Stefanie Jegelka, and Silvio Savarese. Deep metric learning via lifted structured feature embedding. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4004--4012, 2016

    2016

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.