Pith. sign in

REVIEW 5 major objections 5 minor 45 references

HMamba: Hyperbolic Mamba for Sequential Recommendation

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read HMamba runs Mamba's selective state space on a hyperbolic manifold and claims 3-11% accuracy gains over Euclidean and attention-based sequential recommenders at linear-time cost.

desk verdict Clever combination of Mamba and hyperbolic geometry, but the paper's own math makes the curvature-aware mechanism vanish at the stated hyperparameter, so the empirical claim is unattributable as submitted. read the letter →

arxiv 2505.09205 v1 pith:LA4ZRQ3N submitted 2025-05-14 cs.IR

classification cs.IR
keywords sequentialrecommendationhyperbolicgeometrystatespacemodelsMambaLorentzmodelhierarchicalrepresentationlearningrecommendersystemefficiencynext-itemprediction
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 argues that recommendation histories carry a hierarchical structure—genres over items, categories over genres—that Euclidean sequence models flatten, and that Mamba's selective state-space machinery is the right vehicle to exploit that hierarchy if it is run on a hyperbolic manifold instead of in Euclidean space. It proposes HMamba, which lifts item embeddings into the Lorentz model of hyperbolic space, performs curvature-aware discretization of the state-space matrices, mixes features with Möbius gyrovector operations, and scores candidates by hyperbolic distance. On four benchmarks, including ML-1M and three location datasets, the authors report 3-11% improvements over state-of-the-art baselines, with fewer parameters and roughly 3.2x faster training than attention-based recommenders. A hybrid variant that processes in hyperbolic space but scores in Euclidean space performs in between, which the paper reads as evidence that the geometric component, not the extra parameters, drives the gains.

What carries the argument

The load-bearing object is the Lorentz model of hyperbolic space, $H^d_k = \{x \in \mathbb{R}^{d+1} : \langle x,x\rangle_L = -k,\ x_0>0\}$, with curvature parameter $k=1/c$ and Lorentz inner product $\langle x,y\rangle_L = -x_0y_0 + \sum_j x_jy_j$. The construction depends on three operations that keep the state-space model geometrically consistent: the exponential map $\exp_o(v)$ projecting tangent vectors onto the manifold, parallel transport that preserves Lorentz orthogonality during hidden-state updates, and the Möbius gyrovector product used for nonlinear feature mixing. The curvature-aware discretization $\bar A = \exp(\Delta A \odot \operatorname{diag}(\sqrt{|k|},1,\dots,1))$ is the step that nominally imports the geometry into the state-space dynamics, and the complexity analysis attaches a $\log(1/c)$ factor to it. Together these operations let the hidden state evolve along the manifold instead of in a flat ambient space, which is what the paper claims preserves tree-like preference hierarchies.

What would settle it

Run HMamba-Full on ML-1M and Texas with $c=-1$ (negative curvature) and $c=+1$ (positive curvature) under identical hyperparameters and compare HR@10 and NDCG@10; if negative curvature does not improve on positive curvature, or if the positive-curvature run reproduces the reported gains, the claim that hyperbolic geometry drives the results is falsified.

Watch

Extended reading notes

Core claim

HMamba is a sequential recommender whose encoder is a Mamba selective state-space model transplanted onto the Lorentz hyperboloid $H^d_k$ rather than Euclidean space. The pipeline is: zero-pad Euclidean item embeddings to $\mathbb{R}^{d+1}$, apply the exponential map at the origin to obtain points on the hyperboloid, evolve the hidden state through a curvature-aware discretization $\bar A = \exp(\Delta A \odot \operatorname{diag}(\sqrt{|k|},1,\dots,1))$ with parallel transport preserving Lorentz orthogonality, mix features through Möbius gyrovector products, and score the next item by hyperbolic distance to item embeddings lifted onto the same manifold. HMamba-Full performs both representation and scoring in hyperbolic space; HMamba-Half returns to Euclidean space for scoring. The paper's central discovery claim is that this geometric relocation consistently improves next-item prediction over Euclidean Mamba, attention transformers, and the previous hyperbolic attention model HSASRec across four datasets, with the largest margins on the sparser location datasets, while retaining Mamba's linear-time complexity.

Load-bearing premise

The load-bearing premise is that the model is actually operating in a constant-negative-curvature hyperbolic space; the paper defines $c<0$ but fixes $c=1.0$ in its experiments, and if the implemented geometry is not the negative-curvature one, the source of the reported gains is unproven.

Editorial extensions

If this is right

  • If HMamba's reported gains hold, sequential recommenders can reach or exceed attention-based accuracy while scaling linearly with sequence length, making very long user histories practical in production.
  • The Full-versus-Half performance gap implies that scoring in hyperbolic space, not just representing in it, contributes part of the accuracy gain.
  • HMamba would extend selective state-space modeling beyond recommendation to any sequence with intrinsic taxonomy, such as clinical event histories or inventory movements, as the paper itself suggests.
  • The ablation against EMamba indicates that the architectural refinements help even in Euclidean space, so the hyperbolic component is the marginal difference that the paper credits for the improvement.

Reading between the lines

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

  • A fair test of the paper's geometric claim would run HMamba-Full with a genuinely negative curvature constant and compare it against a positive-curvature setting; the reported implementation fixes $c=1.0$, which makes $k=1$ and leaves the Lorentz form with a signature different from the one the paper's theory describes.
  • If the curvature parameter is not actually negative in the reported configuration, the 3-11% gains could be attributable to the Mamba-style architecture rather than to hyperbolic geometry, and the EMamba ablation is consistent with that reading.
  • A testable extension is to sweep $c$ across negative values and check whether gains correlate with $|c|$, which would isolate the curvature mechanism from the architectural mechanism.
  • The complexity argument's $\log(1/c)$ overhead term vanishes at $c=1$, so the paper's complexity comparison implicitly assumes a different curvature setting from the one used in its experiments.
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

5 major / 5 minor

Summary. HMamba proposes to combine Mamba-style selective state space models with hyperbolic geometry for sequential recommendation. The paper defines a Lorentz-model embedding layer, a 'hyperbolic selective state space' with curvature-aware discretization, gyrovector processing, and two model variants (HMamba-Full and HMamba-Half), and reports 3–11% improvements over baselines on ML-1M, New York, California, and Texas datasets. It also provides complexity analysis and several theoretical guarantees (Theorems 2.1–2.4 and Corollary 2.5). The central claim is that adding hyperbolic structure to Mamba yields consistent accuracy gains while retaining linear-time efficiency.

Significance. If the central claim were correct, the paper would introduce a useful architectural direction: extending selective state space models beyond Euclidean geometry for hierarchical sequential data. The idea is timely, and the paper explicitly positions itself against both Transformer-based and Euclidean-Mamba recommenders. The authors also ship a reasonable set of baselines and datasets. However, the paper is not currently publishable because the method's core mathematical formulation is internally inconsistent: the curvature parameter is defined as c < 0 but set to c = 1.0 in the experiments, the discretized transition matrix degenerates to the identity under that setting, the hyperbolic loss function has contradictory sign conventions, and the theoretical guarantees rely on unconstructed constants. The empirical comparison is also underspecified. These are not presentational issues but load-bearing defects that prevent attribution of the reported gains to the claimed hyperbolic mechanism.

major comments (5)
  1. [§2.1/§3.1.4/Algorithm 1] The curvature parameter is internally inconsistent. §2.1 and Table 1 define the hyperbolic space with c < 0 and set k = 1/c, so k < 0. With k < 0, the factors √k appearing in Eqs. (2), (11), and (17) are imaginary, and log(1/c) in Algorithm 1 line 12 is undefined over the reals. In §3.1.4 the paper instead sets c = 1.0, for which log(1/c) = 0; Algorithm 1 line 12 then gives \bar A = exp(ΔA·0) = I, and the K(k) factor in Eq. (13) is the identity. Thus no value of c makes the equations both well-defined and curvature-active: a negative c makes the algorithm's log factor undefined, while c = 1.0 makes the curvature-aware transition inert. The reported 3–11% improvements therefore cannot be attributed to the stated hyperbolic selective state space mechanism.
  2. [§2.5, Eqs. (17)–(18)] The hyperbolic loss is not a valid training objective as written. Eq. (17) defines dL with an explicit negative sign relative to the distance in Eq. (2), making the 'distance' non-positive, and Eq. (18) then sets Lfull = -2k - 2·dL. With the positive distance from Eq. (2), minimizing Lfull would drive the distance to +∞; with the signed quantity from Eq. (17), the loss is proportional to the magnitude but the notation is contradictory. Additionally, Eq. (17) uses '⊗' for elementwise multiplication where the Lorentz inner product is required, and the subscript/superscript placement (e.g., 'ˆ(E_u^h, E_w^h)') is not standard. The objective needs to be re-derived and stated with consistent sign conventions.
  3. [§2.4, Eq. (14)] The formula labeled 'Möbius Gyrovector product' is actually the standard Möbius addition formula for gyrovector spaces, not a product operation. As written, Eq. (14) applies this operation to an SSM output and a SiLU-projected linear transformation, but no definition is given for how these real-valued vectors are mapped to the hyperboloid H^{d+1}_k or how the operation should be interpreted as a layer. This makes the gyrovector processing layer ill-defined and not implementable from the text.
  4. [§2.7, Theorems 2.1–2.4 and Corollary 2.5] The theoretical guarantees are vacuous as stated. The constants C1–C10 are never constructed or bounded, and the proofs never connect them to the model's hyperparameters (A, B, C, Δ, c). Theorem 2.3's proof is a generic covering-number and Rademacher-complexity argument that does not reference the proposed architecture, so the sample-complexity bound cannot be instantiated. Theorem 2.1's proof uses Eq. (23), but Eq. (23) is trivially zero because \bar A = exp(ΔA) by definition in Eq. (5). These results do not establish the claimed sample-efficiency or convergence properties of HMamba.
  5. [§3.1.3/§3.2] The evaluation protocol is underspecified and the headline improvement range is not supported by the reported numbers. No data-split strategy, negative-sampling scheme, or number of random restarts is given, although HR@10, NDCG@10, and MRR@10 all depend on the candidate set used for evaluation. Table 4 marks results with '*' for p-value < 0.01, but no significance test, standard deviation, or number of runs is reported. Moreover, the abstract's '3–11% improvement' claim is inconsistent with Table 4: HMamba-Full improves over the best baseline by only 0.9% HR@10 on New York and 2.5% on California, while on New York HR@10, HMamba-Half ties Mamba4Rec at 0.0544. The experiments need a clearly specified protocol and a presentation of results that matches the claimed range.
minor comments (5)
  1. [Throughout] The manuscript contains numerous typos and formatting errors, including 'frequently' for 'frequently' (Table 1 caption), 'Varnilla Mamba' for 'vanilla Mamba' (§3.2), and 'fraquently' for 'frequently' (§2.2); a thorough proofreading pass is needed.
  2. [Figure 3 caption] The caption says 'the second corresponds to an unspecified region,' but both subplots are labeled 'Texas'; the caption and the figures are inconsistent.
  3. [§3.3/Table 4] The ablation study in Figure 4 reports 'EMamba,' but EMamba is not defined in §3.1.2 and does not appear in Table 4; the variant should be defined and its results reported in a consistent table.
  4. [Eq. (5)] The expression for \bar B, ΔA^{-1} exp(ΔA)·ΔB, is not the standard zero-order-hold discretization; the usual formula is (ΔA)^{-1}(exp(ΔA) − I)ΔB. Please correct the formula or explicitly state the convention.
  5. [Theorems 2.1–2.4] The notation in the theorems is informal: the norm ∥·∥_H, the domain of F, and the relationship between Y and the model output are never precisely defined, and the angle-bracket markup (e.g., in Eq. (17)) appears to be leftover LaTeX formatting.

Circularity Check

1 steps flagged · score 4.0 of 10

The curvature-aware SSM transition collapses to the standard Mamba transition at the reported c=1.0, so the paper's central architectural claim reduces by construction; the external benchmark comparisons remain independent.

  1. self definitional [Algorithm 1 line 12 with Sec. 3.1.4; see also Eq. 13 and Table 2]
    "¯A← exp(ΔA· log(1/c)) ⊳ Curvature-adjusted ... The curvature parameter (c) governing the underlying hyperbolic space was set at 1.0."

    Substituting the paper's own reported hyperparameter c=1.0 into its own transition definition gives log(1/c)=log(1)=0, hence ¯A=exp(0·ΔA)=I. Eq. 13's curvature factor K(k)=diag(sqrt(|k|),1,...,1) is also the identity at k=1/c=1. The 'curvature-adjusted' hyperbolic selective state space therefore reduces by construction to the ordinary Mamba transition at the setting actually evaluated, and the curvature-dependent log(1/c) complexity factor in Table 2 vanishes. The claimed 3-11% gains are attributed to a curvature-aware mechanism that, under the paper's own equations and hyperparameter, is inert; the novel mechanism is definitionally equivalent to its Euclidean Mamba input at this setting.

full rationale

The main accuracy claim is not a fitted prediction: Table 4 compares against external datasets (ML-1M, New York, California, Texas) and published baselines (BPR-MF, Caser, NARM, GRU4Rec, SASRec, BERT4Rec, LRURec, SR-GNN, Mamba4Rec, HSASRec), so the reported 3-11% improvements are externally anchored and not constants fitted to a target and then renamed as predictions. The theoretical results (Theorems 2.1-2.5) are generic covering-number, Rademacher, and SGD arguments with constants C1-C10 never constructed; they are vacuous and unsupported, but vacuousness is a missing-support/correctness problem rather than a circular derivation. Related-work self-citations, such as [42] with an overlapping author, are motivational and do not carry the argument. The one genuine reduction by construction is the curvature parameter: Sec. 3.1.4 fixes c=1.0 while Algorithm 1 defines the 'curvature-adjusted' update as exp(ΔA·log(1/c)), making the adjustment identity, and Eq. 13's K(k) is identity at k=1. Thus the central architectural novelty 'hyperbolic selective state space' degenerates to the standard Mamba transition at the deployed setting, so the mechanism credited for the gains is not the mechanism defined. Because the empirical evaluation itself remains independent and externally comparable, the score is 4 rather than 6 or higher.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central model relies on the curvature parameter c, the embedding dimension, and the discretization timestep as free choices, none of which are given a principled derivation. The hyperbolic operations are built on standard geometry axioms plus an ad hoc curvature-normalized discretization and a mislabeled gyrovector product. The theoretical section adds unspecified constants rather than testable constraints.

free parameters (3)
  • curvature parameter c = 1.0
    Chosen in Sec. 3.1.4 even though Eq. 1 defines c < 0. With k = 1/c = 1, the space is not the stated negative-curvature hyperbolic space, and the log(1/c) factor in Table 2 vanishes.
  • embedding dimension d = 32
    Set in Sec. 3.1.4. All experimental results depend on this choice, and no sensitivity analysis is shown across datasets.
  • discretization timestep Δ
    Used in Eq. 5 and Eq. 13 but never given a value or schedule, so the state space discretization is not documented.
assumptions (5)
  • standard math Lorentz model definitions, exponential map, and parallel transport as established in hyperbolic representation learning literature
    Background geometry taken from cited works such as Nickel & Kiela (2018) and Sun et al. (2021); used in Sec. 2.3 without proof.
  • domain assumption Recommendation data exhibits intrinsic hierarchical structure that Euclidean embeddings cannot represent efficiently
    Motivates the entire approach in Sec. 1 and Fig. 1, but no quantitative evidence on the four datasets is provided.
  • ad hoc to paper Curvature-normalized discretization \bar{A} = exp(ΔA ⊙ K(k)) with K(k) = diag(sqrt(|k|), 1, ..., 1) preserves metric properties and selectivity
    Introduced in Eq. 13 with no derivation; this is the core of the claimed 'hyperbolic selective state space' and is not a standard construction.
  • ad hoc to paper The formula in Eq. 14 for the Möbius gyrovector product is actually the Möbius addition formula
    The operations are mislabeled, and no source is cited for the specific product form used in the layer.
  • ad hoc to paper Theorems 2.1-2.4 assume bounding constants C1-C10 that are independent of sequence length and dimension
    The constants are never constructed, and the bounds are not shown to be finite for the actual HMamba computation, so the theoretical claims are not operational.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HMamba: Hyperbolic Mamba for Sequential Recommendation." pith.science (2026). https://pith.science/paper/LA4ZRQ3N

@misc{pith2026250509205,
  author       = {Pith},
  title        = {Pith review of: HMamba: Hyperbolic Mamba for Sequential Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LA4ZRQ3N}},
  note         = {Machine review of arXiv:2505.09205}
}
read the original abstract

Sequential recommendation systems have become a cornerstone of personalized services, adept at modeling the temporal evolution of user preferences by capturing dynamic interaction sequences. Existing approaches predominantly rely on traditional models, including RNNs and Transformers. Despite their success in local pattern recognition, Transformer-based methods suffer from quadratic computational complexity and a tendency toward superficial attention patterns, limiting their ability to infer enduring preference hierarchies in sequential recommendation data. Recent advances in Mamba-based sequential models introduce linear-time efficiency but remain constrained by Euclidean geometry, failing to leverage the intrinsic hyperbolic structure of recommendation data. To bridge this gap, we propose Hyperbolic Mamba, a novel architecture that unifies the efficiency of Mamba's selective state space mechanism with hyperbolic geometry's hierarchical representational power. Our framework introduces (1) a hyperbolic selective state space that maintains curvature-aware sequence modeling and (2) stabilized Riemannian operations to enable scalable training. Experiments across four benchmarks demonstrate that Hyperbolic Mamba achieves 3-11% improvement while retaining Mamba's linear-time efficiency, enabling real-world deployment. This work establishes a new paradigm for efficient, hierarchy-aware sequential modeling.

Figures

Figures reproduced from arXiv: 2505.09205 by the authors.

Figure 1
Figure 1. Hyperbolic and Euclidean embeddings of ML-1M hierarchy structure, showing (1) Genre (4 kinds: Action, Comedy, Drama, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The diagram illustrates the configuration of the [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Performance comparison of baselines on different groups of data [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Ablation Study on Data with the Sequence Length [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]
Figure 5
Figure 5. Figure 5: GPU cost comparison of SASRec, BERT4Rec, LRURec, Mamba4Rec, [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 6
Figure 6. Figure 6: Training time and testing time comparison of SASRec, BERT4Rec, LRURec, Mamba4Rec, [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Case study of embeddings of HMamba-Full and Mamba4Rec based on New York and California datasets resources at 6.77GB GPU memory and 50.77s per epoch. Interestingly, some larger configurations demonstrate comparable or even faster training times than medium-sized ones, s…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 30 canonical work pages

  1. [1]

    Tesfaye Fenta Boka, Zhendong Niu, and Rama Bastola Neupane. 2024. A survey of sequential recommendation systems: Techniques, evaluation, and future directions.Information Systems(2024), 102427. Manuscript submitted to ACM HMamba: Hyperbolic Mamba for Sequential Recommendation 23

  2. [2]

    Benjamin Paul Chamberlain, Stephen R Hardwick, David R Wardrope, Fabon Dzogang, Fabio Daolio, and Saúl Vargas. 2019. Scalable hyperbolic recommender systems.arXiv preprint arXiv:1902.08648(2019)

  3. [3]

    Ines Chami, Rex Ying, Christopher Ré, and Jure Leskovec. 2020. Hyperbolic Graph Neural Networks.NeurIPS(2020)

  4. [4]

    Qiang Cui, Shu Wu, Qiang Liu, Wen Zhong, and Liang Wang. 2018. MV-RNN: A multi-view recurrent neural network for sequential recommendation. IEEE Transactions on Knowledge and Data Engineering32, 2 (2018), 317–331

  5. [5]

    Gabriel de Souza Pereira Moreira, Sara Rabhi, Jeong Min Lee, Ronay Ak, and Even Oldridge. 2021. Transformers4rec: Bridging the gap between nlp and sequential/session-based recommendation. InProceedings of the 15th ACM conference on recommender systems. 143–153

  6. [6]

    Ziwei Fan, Zhiwei Liu, Jiawei Zhang, Yun Xiong, Lei Zheng, and Philip S Yu. 2021. Continuous-time sequential recommendation with temporal graph collaborative transformer. InProceedings of the 30th ACM international conference on information & knowledge management. 433–442

  7. [7]

    Evgeny Frolov, Lina Bashaeva, Leyla Mirvakhabova, and Ivan Oseledets. [n.d.]. Hyperbolic Embeddings in Sequential Self-Attention for Improved Next-Item Recommendations. ([n.d.])

  8. [8]

    Octavian Ganea, Gary Bécigneul, and Thomas Hofmann. 2018. Hyperbolic Neural Networks.Advances in Neural Information Processing Systems31 (2018). https://proceedings.neurips.cc/paper/2018/file/ffeed84c7cb1ae957bfc4d9000fa9f38-Paper.pdf

Show all 45 references
  1. [9]

    Albert Gu and Tri Dao. 2023. Mamba: Linear-time sequence modeling with selective state spaces.arXiv preprint arXiv:2312.00752(2023)

  2. [10]

    Poincaréheterogeneous graph neural networks for sequential recommendation.ACM Transactions on Information Systems41, 3 (2023), 1–26

    NaichengGuo,XiaoleiLiu,ShaoshuaiLi,QiongxuMa,KaixinGao,BingHan,LinZheng,ShengGuo,andXiaoboGuo.2023. Poincaréheterogeneous graph neural networks for sequential recommendation.ACM Transactions on Information Systems41, 3 (2023), 1–26

  3. [11]

    B Hidasi. 2015. Session-based Recommendations with Recurrent Neural Networks.arXiv preprint arXiv:1511.06939(2015)

  4. [12]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recommendation. In2018 IEEE international conference on data mining (ICDM). IEEE, 197–206

  5. [13]

    Jiong Li and Pratik Gajane. 2023. Curiosity-driven exploration in sparse-reward multi-agent reinforcement learning.arXiv preprint arXiv:2302.10825 (2023)

  6. [14]

    Jing Li, Pengjie Ren, Zhumin Chen, Zhaochun Ren, Tao Lian, and Jun Ma. 2017. Neural attentive session-based recommendation. InProceedings of the 2017 ACM on Conference on Information and Knowledge Management. 1419–1428

  7. [15]

    Jiacheng Li, Jingbo Shang, and Julian McAuley. 2022. Uctopic: Unsupervised contrastive learning for phrase representations and topic mining.arXiv preprint arXiv:2202.13469(2022)

  8. [16]

    Yicong Li, Hongxu Chen, Xiangguo Sun, Zhenchao Sun, Lin Li, Lizhen Cui, Philip S Yu, and Guandong Xu. 2021. Hyperbolic hypergraphs for sequential recommendation. InProceedings of the 30th ACM international conference on information & knowledge management. 988–997

  9. [17]

    Chengkai Liu, Jianghao Lin, Jianling Wang, Hanzhou Liu, and James Caverlee. 2024. Mamba4rec: Towards efficient sequential recommendation with selective state space models.arXiv preprint arXiv:2403.03900(2024)

  10. [18]

    Larry R Medsker, Lakhmi Jain, et al. 2001. Recurrent neural networks.Design and Applications5, 64-67 (2001), 2

  11. [19]

    Leyla Mirvakhabova, Evgeny Frolov, Valentin Khrulkov, Ivan Oseledets, and Alexander Tuzhilin. 2020. Performance of hyperbolic geometry models on top-n recommendation tasks. InProceedings of the 14th ACM Conference on Recommender Systems. 527–532

  12. [20]

    Gal Mishne, Zhengchao Wan, Yusu Wang, and Sheng Yang. 2023. The numerical stability of hyperbolic representation learning. InInternational Conference on Machine Learning. PMLR, 24925–24949

  13. [21]

    Poincaréembeddingsforlearninghierarchicalrepresentations

    MaximillianNickelandDouweKiela.2017. Poincaréembeddingsforlearninghierarchicalrepresentations. Advancesinneuralinformationprocessing systems30 (2017)

  14. [22]

    Maximillian Nickel and Douwe Kiela. 2018. Learning continuous hierarchies in the lorentz model of hyperbolic geometry. InInternational conference on machine learning. PMLR, 3779–3788

  15. [23]

    Wei Peng, Tuomas Varanka, Abdelrahman Mostafa, Henglin Shi, and Guoying Zhao. 2021. Hyperbolic deep neural networks: A survey.IEEE Transactions on pattern analysis and machine intelligence44, 12 (2021), 10023–10044

  16. [24]

    Haohao Qu, Yifeng Zhang, Liangbo Ning, Wenqi Fan, and Qing Li. 2024. Ssd4rec: a structured state space duality model for efficient sequential recommendation. arXiv preprint arXiv:2409.01192(2024)

  17. [25]

    Massimo Quadrana, Paolo Cremonesi, and Dietmar Jannach. 2018. Sequence-aware recommender systems.ACM computing surveys (CSUR)51, 4 (2018), 1–36

  18. [26]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2012. BPR: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618(2012)

  19. [27]

    Alex Sherstinsky. 2020. Fundamentals of recurrent neural network (RNN) and long short-term memory (LSTM) network.Physica D: Nonlinear Phenomena 404 (2020), 132306

  20. [28]

    Jinzhao Su and Zhenhua Huang. 2024. Mlsa4rec: Mamba combined with low-rank decomposed self-attention for sequential recommendation.arXiv preprint arXiv:2407.13135(2024)

  21. [29]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang. 2019. BERT4Rec: Sequential recommendation with bidirectional encoder representations from transformer. InProceedings of the 28th ACM international conference on information and knowledge management. 1441–1450

  22. [30]

    Jianing Sun, Zhaoyue Cheng, Saba Zuberi, Felipe Pérez, and Maksims Volkovs. 2021. Hgcf: Hyperbolic graph convolution networks for collaborative filtering. InProceedings of the Web Conference 2021. 593–601. Manuscript submitted to ACM 24 Qianru Zhang, Honggang Wen, Wei Yuan, Cr...

  23. [31]

    Jiaxi Tang and Ke Wang. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. InProceedings of the eleventh ACM international conference on web search and data mining. 565–573

  24. [32]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

  25. [33]

    Lucas Vinh Tran, Yi Tay, Shuai Zhang, Gao Cong, and Xiaoli Li. 2020. Hyperml: A boosting metric learning approach in hyperbolic space for recommender systems. InProceedings of the 13th international conference on web search and data mining. 609–617

  26. [34]

    Shoujin Wang, Liang Hu, Yan Wang, Longbing Cao, Quan Z Sheng, and Mehmet Orgun. 2019. Sequential recommender systems: challenges, progress and prospects.arXiv preprint arXiv:2001.04830(2019)

  27. [35]

    Shu Wu, Yuyuan Tang, Yanqiao Zhu, Liang Wang, Xing Xie, and Tieniu Tan. 2019. Session-based recommendation with graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, Vol. 33. 346–353

  28. [36]

    Sheng, Zhiming Cui, Xiaofang Zhou, and Hui Xiong

    Chengfeng Xu, Pengpeng Zhao, Yanchi Liu, Jiajie Xu, Victor S Sheng S. Sheng, Zhiming Cui, Xiaofang Zhou, and Hui Xiong. 2019. Recurrent convolutional neural network for sequential recommendation. InThe world wide web conference. 3398–3404

  29. [37]

    An Yan, Zhankui He, Jiacheng Li, Tianyang Zhang, and Julian McAuley. 2023. Personalized showcases: Generating multi-modal explanations for recommendations. InProceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2251–2255

  30. [38]

    Jiyuan Yang, Yuanzi Li, Jingyu Zhao, Hanbing Wang, Muyang Ma, Jun Ma, Zhaochun Ren, Mengqi Zhang, Xin Xin, Zhumin Chen, and Pengjie Ren

  31. [39]

    Hypformer:ExploringEfficientHyperbolicTransformer Fully in Hyperbolic Space

    MenglinYang,HarshitVerma,DelvinCeZhang,JiahongLiu,IrwinKing,andRexYing.2024. Hypformer:ExploringEfficientHyperbolicTransformer Fully in Hyperbolic Space. InProceedings of the 2024 ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

  32. [40]

    Sequentialrecommendationwithlatentrelations based on large language model

    ShenghaoYang,WeizhiMa,PeijieSun,QingyaoAi,YiqunLiu,MingchenCai,andMinZhang.2024. Sequentialrecommendationwithlatentrelations based on large language model. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 335–344

  33. [41]

    Hongzhi Yin, Yizhou Sun, Bin Cui, Zhiting Hu, and Ling Chen. 2013. Lcars: a location-content-aware recommender system. InProceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining. 221–229

  34. [42]

    Yonghong Yu, Aoran Zhang, Li Zhang, Rong Gao, Shang Gao, and Hongzhi Yin. 2024. Hyperbolic translation-based sequential recommendation. IEEE Transactions on Computational Social Systems(2024)

  35. [43]

    Zhenrui Yue, Yueqi Wang, Zhankui He, Huimin Zeng, Julian McAuley, and Dong Wang. 2024. Linear recurrent units for sequential recommendation. InProceedings of the 17th ACM International Conference on Web Search and Data Mining. 930–938

  36. [44]

    Qianru Zhang, Lianghao Xia, Xuheng Cai, Siu-Ming Yiu, Chao Huang, and Christian S Jensen. 2024. Graph augmentation for recommendation. In 2024 IEEE 40th International Conference on Data Engineering (ICDE). IEEE, 557–569. Manuscript submitted to ACM

  37. [2024]

    arXiv:2403.16371 [cs.IR]

    Uncovering Selective State Space Model’s Capabilities in Lifelong Sequential Recommendation. arXiv:2403.16371 [cs.IR]

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.