{"id":"6d457043-9a83-445b-b06e-8bd8225e2c1e","arxiv_id":"2412.17197","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Q-LIME π encodes binary feature vectors as qubit states and flips bits to estimate feature contributions, but the quantum encoding is decorative and the method reduces to classical leave-one-out attribution.","lead":"This paper wraps LIME's feature-removal trick in quantum language, encoding binary features as qubit rotations and flipping them with Pauli-X gates, and reports similar top features with faster runtimes on tiny IMDb subsets. Generalists might read it as a quantum-inspired explainability pitch, but the quantum part is a classical simulation and the algorithm reduces to a standard leave-one-out ablation.","discovery_kind":"incremental","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Algorithm 1's central flip is a no-op: X|+>=|+>, so applying Pauli-X to an encoded 1-feature does not change it; Section 4.2 silently substitutes a different classical operation.","rationale":"The reader's weakest assumption is exactly where I landed: X|+>=|+> invalidates Algorithm 1. I checked the encoding and perturbation definitions in Sections 3 and 4.1, and the discrepancy with Section 4.2's implementation details; the math is unambiguous. This is not a dispute about whether quantum superposition could be useful in some future version; it is an internal inconsistency in the presented algorithm. Because the experimental section depends on the implementation's unstated classical substitute, the paper's central claim about quantum flips is unsupported. I agree with the reader's REJECT verdict, so no adjustment is needed. I did not find a separate more basic flaw; the Pauli-X no-op is the load-bearing one.","tokens_in":5616,"tokens_out":5093,"duration_ms":47607,"concrete_test":"Analytically compute X Ry(pi/2)|0>: since Ry(pi/2)|0>=|+> and X|+>=|+>, it returns |+>, proving the target bit is unchanged. To see the consequence, run Algorithm 1 literally on a PennyLane simulator for a fixed IMDb instance: for every k with xk=1, apply qml.PauliX to the qubit prepared by qml.RY(pi/2, wires=k) and measure; if the measured bit is always 1, the claimed 1->0 flip cannot occur, so the reported results must have come from Section 4.2's substitution rather than Algorithm 1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Definition 4 encodes xi=1 as Ry(pi/2)|0> = (|0>+|1>)/sqrt2 = |+>. Section 4.1 and Algorithm 1 line 5 then claim Pauli-X toggles this bit from 1 to 0, but X|+>=|+>, so the operation is identity on the target qubit. A literal execution of Algorithm 1 can never produce the 1->0 perturbations on which Definition 3, the Delta f_k scores, and the experimental comparisons depend. The implementation (Section 4.2) avoids this by setting the RY angle to 0 when xi=1; that is an unstated classical bit-removal, not X_k, and therefore the reported overlaps and runtimes do not validate the described quantum algorithm. Additionally, after such an angle reset the remaining 1-features are still in |+>, so a computational-basis measurement yields a random subset of the other present features; Delta f_k is then f(x)-f(random-subset), not f(x)-f(x with only xk removed) as in Definition 3. The central claim therefore rests on a mechanism that is either a no-op or an undocumented classical substitute.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Q-LIME π, a quantum-inspired extension of LIME for binary feature vectors. The method encodes a binary input via single-qubit Ry rotations (Definition 4), claims that applying a Pauli-X gate to the k-th qubit flips the encoded feature from 1 to 0 (Section 4.1), computes Δf_k = f(x) − f(x_perturbed,k), and fits a linear surrogate model (Algorithm 1). Experiments on a small IMDb subset compare top-5 feature overlap with classical LIME and report runtime reductions. The central mechanism, however, is mathematically incorrect as stated: with the encoding used, Pauli-X is a no-op on the feature-present state. The implementation described in Section 4.2 silently replaces the Pauli-X flip with a classical angle reset, so the reported experiments do not validate the quantum algorithm presented in the paper.","tokens_in":5889,"tokens_out":4298,"duration_ms":39396,"significance":"A faster, faithful LIME-like explainer for low-dimensional binary feature spaces would be a useful contribution, and the paper addresses a real limitation of LIME's perturbation cost. The authors also provide a public code repository, which supports reproducibility. However, the core quantum perturbation operation is invalid under the paper's own encoding, and the implementation deviates from the algorithm without explanation. In addition, the experimental evaluation is too limited (5 unseeded test instances, no error bars, selected configurations) to support the central claims of 'near-identical top-feature rankings' and systematic runtime improvement. The claimed quantum-inspired efficiency gain is essentially by construction because Q-LIME π performs n forward passes rather than LIME's ~300 perturbations, and the paper does not demonstrate that this reduced evaluation preserves explanation fidelity. As presented, the contributions are not established.","major_comments":[{"comment":"The central perturbation operation is a no-op. Definition 4 encodes xi=1 as Ry(π/2)|0> = |+>. The Pauli-X gate satisfies X|+> = |+>, so applying X_k to the state prepared by Definition 4 leaves the k-th qubit unchanged. A literal execution of Algorithm 1 therefore cannot produce the 1→0 flip on which Definition 3 and all subsequent Δf_k computations depend. The text's justification that 'X_k toggles |0> ↔ |1>' refers to computational-basis states, not to the |+> states used in the encoding.","section":"Section 4.1 and Algorithm 1 (line 5)"},{"comment":"The implementation silently substitutes a different operation for the Pauli-X flip. The text states that 'if xi = 1, we set the corresponding RY angle to 0' and then measures. This is a classical bit-removal, not the X_k gate of Algorithm 1. Because the experiments use this undocumented substitution, the reported overlaps and runtimes do not validate the quantum algorithm proposed in the abstract and Section 4.1. The gap between the algorithm and its implementation is load-bearing for every experimental claim in the paper.","section":"Section 4.2, Implementation Details"},{"comment":"The evaluation is statistically insufficient to support the main claims. Only 5 random test instances are used per configuration, with no mention of random seeds, repeated trials, or error bars. The reported accuracy values range from 0.44 to 0.65 across configurations, and Table 1 presents a selected subset of the full parameter grid. The overlap values (2.8–5.0) also have no variance estimates. The conclusion that Q-LIME π 'often achieves near-identical top-feature rankings' is not supported by this limited evidence.","section":"Section 5, Table 1"},{"comment":"Even if the angle-reset implementation is taken as the intended method, the measured perturbed state does not correspond to Definition 3. After setting the RY angle of the selected feature to 0, all other features that are present remain in |+>, and a computational-basis measurement yields a random subset of those other present features. Thus Δf_k = f(x) − f(random subset), not f(x) − f(x with only xk removed). The surrogate model in Algorithm 1 therefore does not measure the claimed feature contribution, and the comparison with LIME's deterministic removal strategy is not meaningful without accounting for this randomness.","section":"Definition 3 vs. Section 4.2 measurement"}],"minor_comments":[{"comment":"The sentence 'X_k toggles |0> ↔ |1>' is correct for computational-basis states but misleading here, because the encoded state is |+>; please specify that the Pauli-X operation is applied to the computational basis after the Ry encoding, or revise the encoding so that the claimed flip is actually implemented.","section":"Section 4.1"},{"comment":"The 'shots' column lists None and 100, but the algorithm says 'measure' and 'sample one outcome.' With shots=None (analytic mode), PennyLane returns probabilities rather than a single bitstring, so it is unclear what 'sample one outcome' means in that case; please clarify.","section":"Table 1 and Section 4.2"},{"comment":"The three example reviews appear to be selected rather than randomly drawn, but the selection criterion is not stated. Please explain how these examples were chosen and whether they are representative of the average overlap reported in Table 1.","section":"Table 2"},{"comment":"The paper refers to 'Pira and Ferrie's Q-LIME' but the cited title is 'On the interpretability of quantum neural networks.' Please clarify whether this is the same method or cite a separate Q-LIME paper if one exists.","section":"Related Work, reference [7]"},{"comment":"The GitHub URL contains a space ('https://github. com/nelabdiel/qlime'); also, for reproducibility, please provide a versioned release or archive DOI rather than a link to a possibly changing repository.","section":"Data Availability"}],"recommendation":"reject","confidential_remarks":"The paper's core algorithmic claim is invalid under its own encoding, and the implementation deviates from the algorithm without explanation. The experimental design is far too weak to support the stated conclusions. I would be willing to consider a revised version that fixes the perturbation semantics, clearly separates the classical implementation from the quantum-inspired description, and provides a proper statistical evaluation with seeds and error bars. As it stands, the manuscript does not meet the standards for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe one thing you should know before reading: the central mechanism is mathematically wrong. Definition 4 encodes a present feature as |+> = Ry(π/2)|0>, and Algorithm 1 then claims that applying Pauli-X toggles that bit from 1 to 0. But X|+> = |+>, so that operation is the identity. The implementation in Section 4.2 silently sidesteps this by setting the rotation angle to 0 for the flipped bit—which is a classical bit removal, not a Pauli-X flip. Worse, with the other present features still in |+>, a computational-basis measurement collapses them randomly, so the computed Δf_k is f(x) minus f(random subset of the other 1-features), not f(x) minus f(x with only x_k removed) as Definition 3 requires. The paper's core computation, as both described and implemented, does not do what it claims.\n\nThat said, the paper has some virtues. It is clearly written, cites the relevant literature including the existing Q-LIME of Pira and Ferrie, and honestly distinguishes itself as \"quantum-inspired\" rather than claiming immediate hardware advantage. The GitHub repository is a plus. The experimental section is transparent about the small scale and the exponential classical simulation cost, and the runtime comparison is honestly framed as a proof-of-concept.\n\nBut the soft spots are load-bearing. The intended method reduces to leave-one-out feature ablation, a standard baseline that predates LIME, and the quantum encoding adds nothing functionally—it is a relabeling of binary features. The runtime advantage is by construction: Q-LIME π makes n forward passes while LIME makes ~300 perturbations; of course it is faster. The reported overlap with LIME is not evidence of correctness unless Δf_k is shown to approximate LIME's surrogate weights, and no such analysis is given. The experiments use five unseeded instances per configuration, no error bars, and only a subset of configurations. These would be minor issues in an otherwise sound paper, but here they compound the central flaw.\n\nWho is this paper for? A reader curious about quantum-inspired XAI might skim it as an example of a common pitfall, but it is not a reliable contribution. It does not deserve a serious referee: the basic quantum error is disqualifying, and the claimed novelty and evidence do not hold up.\n\nMy recommendation: pass on it. If you engage, do so as a teaching example of why quantum-inspired methods must be checked against the actual math.","headline":"The central quantum flip is a no-op (X|+>=|+>), and the implementation substitutes a different classical operation, so the paper's main claim rests on a mechanism that does not work as described.","tokens_in":6353,"tokens_out":4382,"would_cite":false,"duration_ms":35244,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Q-LIME π claims to reproduce LIME's top-feature rankings on binary text data while cutting runtime, by encoding features in a quantum state and flipping present bits with Pauli-X operations.","keywords":["Explainable AI","Quantum Machine Learning","Q-LIME π","Qubit-Based Encoding","Quantum-Classical Hybrid","LIME","Feature Perturbation","Local Explanations"],"falsifier":"In a simulator or on real hardware, prepare a qubit with $R_y(\\pi/2)|0\\rangle$, apply the Pauli-X gate, and measure: the outcome is $|+\\rangle$ (feature still present), not $|0\\rangle$ (feature removed), which directly refutes the toggling claim in Algorithm 1.","tokens_in":5421,"feed_emoji":"⚛️","tokens_out":4432,"duration_ms":39220,"temperature":0.7,"pith_summary":"The paper proposes a quantum-inspired variant of LIME that encodes a binary feature vector into a product of single-qubit rotations and expresses feature removal as a bit flip on the encoded state. The central claim is that this approach, called Q-LIME π, often reproduces classical LIME's top-5 feature rankings on IMDb sentiment reviews while reducing runtime by 39% to 98% in small- to moderate-dimensional feature spaces. A sympathetic reader would care because local interpretability methods like LIME become expensive as feature counts grow, and a faster surrogate that preserves LIME's behavior could make explanations more practical in constrained settings.","feed_headline":"Q-LIME π: LIME's top features, up to 98% faster","feed_subtitle":"Encoding binary features as qubit rotations reproduces LIME's top-5 rankings on IMDb sentiment with big runtime savings.","key_machinery":"The central object is the quantum state encoding $|\\psi\\rangle = \\bigotimes_{i=1}^n R_y(\\theta_i)|0\\rangle$ with $\\theta_i = \\pi/2$ when $x_i = 1$ and $\\theta_i = 0$ when $x_i = 0$, so present features sit in the $|+\\rangle$ state. The perturbation step applies the Pauli-X gate $X_k$ to qubit $k$, intending to toggle the $k$-th feature from 1 to 0; the measurement outcome is then used to compute $\\Delta f_k = f(x) - f(x_{\\text{perturbed},k})$ and to fit a linear surrogate $g(x) = \\sum_k \\Delta f_k x_k$ that assigns importance scores.","core_discovery":"On its own terms, the paper's discovery is that a quantum-inspired perturbation protocol—encoding only the active features of a binary vector via $R_y(\\pi/2)$ rotations, applying a Pauli-X gate to each present feature to toggle it off, measuring the resulting bitstring, and feeding it to a logistic-regression classifier—yields feature-importance scores closely aligned with classical LIME. Across configurations, the average overlap with LIME's top-5 features is about 3.8 out of 5, with per-instance overlaps reaching 5 out of 5, and the measured runtime is consistently lower than LIME's. The author interprets this as evidence that superposition and interference can be leveraged to explore local neighborhoods more efficiently, at least in low-dimensional settings.","pith_inferences":["The reported speedup likely stems from sampling a single perturbed state per feature rather than from quantum parallelism, since the simulator is classical and each flip is processed sequentially.","A direct comparison against a classical baseline that removes each present feature one at a time would reveal whether the quantum encoding contributes anything beyond a cheap deterministic sampling strategy.","If the toggling operation were corrected, the method would generalize cleanly to any binary tabular dataset, but the high-dimensional ambitions would require encoding or sampling schemes that avoid exponential state-vector simulation.","The average overlap of 3.8 out of 5 top features is modest in absolute terms, suggesting that LIME's random perturbations and Q-LIME π's deterministic flips may diverge on borderline tokens; measuring agreement over many instances would clarify how often near-identical rankings actually occur."],"forward_implications":["If the claim holds, Q-LIME π can serve as a drop-in replacement for LIME in binary-feature, low-dimensional settings, delivering comparable explanations at lower computational cost.","The bit-flip logic extends directly to the 0→1 direction, allowing the same framework to model feature addition as well as removal.","The encoding suggests a path toward genuine quantum parallelism, where multiple perturbation states could be prepared and measured in superposition rather than one at a time.","The method inherits LIME's assumptions of local linearity and feature independence, so any limitations of LIME along those dimensions carry over unchanged.","Runtime advantages in practice depend on the cost of state preparation and measurement; the reported gains hold for up to about 15 features on a classical simulator."],"supporting_citations":[{"why":"Defines LIME, the baseline method whose perturbation and removal strategy Q-LIME π emulates.","marker":"[8]"},{"why":"Supplies the IMDb sentiment dataset used in all experiments.","marker":"[6]"},{"why":"PennyLane is the simulation framework used to encode states, apply gates, and obtain measurement outcomes.","marker":"[2]"},{"why":"Motivates the quantum machine learning angle and the claimed efficiency benefits of superposition and interference.","marker":"[3]"},{"why":"Prior Q-LIME work that the paper distinguishes from its own binary-feature, classical-LIME-bridging focus.","marker":"[7]"}],"fun_headline_variants":["Quantum-inspired LIME matches key features at lower cost","Q-LIME π speeds up LIME explanations","Flipping bits quantum-style reproduces LIME's top picks faster","Quantum encoding matches LIME's top-5 features in less time"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method assumes that applying a Pauli-X gate to a qubit in the $|+\\rangle$ state removes a present feature by flipping it to $|0\\rangle$, but in fact $X|+\\rangle = |+\\rangle$, so the operation described in the algorithm does nothing.","fun_headline_variants_meta":{"raw":{"variants":["Quantum-inspired LIME matches key features at lower cost","Q-LIME π speeds up LIME explanations","Flipping bits quantum-style reproduces LIME's top picks faster","Quantum encoding matches LIME's top-5 features in less time"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000642,"raw_usage":{"total_tokens":2941,"prompt_tokens":917,"completion_tokens":2024,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":533,"completion_tokens_details":{"reasoning_tokens":1955}},"tokens_in":533,"tokens_out":2024,"duration_ms":15708,"temperature":1.0,"reasoning_tokens":1955,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:42:18.765138+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In a simulator or on real hardware, prepare a qubit with $R_y(\\pi/2)|0\\rangle$, apply the Pauli-X gate, and measure: the outcome is $|+\\rangle$ (feature still present), not $|0\\rangle$ (feature removed), which directly refutes the toggling claim in Algorithm 1.","supporting_citations":[{"cited_title":"Why Should I Trust You?","cited_arxiv_id":null,"evidence_quote":"Defines LIME, the baseline method whose perturbation and removal strategy Q-LIME π emulates."},{"cited_title":"Learn- ing word vectors for sentiment analysis","cited_arxiv_id":null,"evidence_quote":"Supplies the IMDb sentiment dataset used in all experiments."},{"cited_title":"Quantum machine lear ning","cited_arxiv_id":null,"evidence_quote":"Motivates the quantum machine learning angle and the claimed efficiency benefits of superposition and interference."},{"cited_title":"On the interpretability of quantum neural ne tworks","cited_arxiv_id":null,"evidence_quote":"Prior Q-LIME work that the paper distinguishes from its own binary-feature, classical-LIME-bridging focus."}],"review_version":1}