paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

arXiv

Evaluating Counterfactual Sensitivity to Patient Information in Medication-Safety Reasoning

Zhitian Hou, Yuhang Liu, Pengkai Wang, Zeyu Liu, Guanghao Zhu, Zheng Liu, Shuo Cai, Congkai Xie, Zhijie Sang, Kun Zeng, Hongxia Yang

Applying a valid medication-safety rule when its patient-specific conditions are not met can produce an incorrect decision. Existing medical evaluations largely use isolated and fixed scenarios. A model may therefore answer correctly by recalling a drug-risk association without showing that it used patient information to decide whether the rule applies. To address this gap, we introduce MedPIC-Bench, a benchmark of source-verifiable recommendations and expert-validated questions for patient-specific medication-safety reasoning. It combines guideline-following questions with paired counterfactual questions in which a controlled change in patient information changes whether a rule applies. The benchmark contains 467 questions annotated along six clinical and reasoning dimensions. Across 28 medical-specific, general, and proprietary LLMs, every model performs worse on counterfactual questions, with mean accuracy falling from 63.6\% to 45.1\%. Models perform well when an explicit patient attribute directly signals a familiar contraindication, but struggle when patient information must narrow or withdraw a safety warning. Model rationales often acknowledge the changed patient information, yet the final answers retain the previous safety judgment. This vulnerability persists among medical-specific LLMs, whose average CF performance trails that of general LLMs. MedPIC-Bench therefore makes conditional rule application measurable and highlights the limitations of static medication-safety accuracy for assessing patient-specific reliability.

arXiv

UrbanAgent: A Tool-Augmented Agent for Cross-System Urban Tasks

Jiayu Cao, Xingyuan Zeng, feiyu Li, Zhijing Huang, Xujie Yuan, Rongxiang Chen, Shimin Di, Libin Zheng, Jian Yin

Modern cities rely on an increasing number of digital services to operate, but residents' daily needs are still difficult to meet. Services are fragmented and have little interoperability, placing a heavy operational burden on users. Existing digital platforms, urban foundation models, and intelligent assistants each address only isolated aspects of an urban task. But they struggle to reliably convert complex natural-language requests into executable cross-system workflows. We propose Urban-Agent, a tool-augmented agent framework for cross-system urban tasks. It couples the cognitive and reasoning capabilities of a large language model with a tool-set supporting code execution, API calls, and Model Context Protocol. Through one adaptive closed loop, it clarifies missing information before acting, grounds tool use in live observations, and aligns the final response with observed evidence and task constraints. To address the evaluation gap, we introduce Urban-Eval, a benchmark specifically designed for cross-system urban request. Unlike prior benchmarks that assess either general tool use or urban knowledge and reasoning, Urban-Eval evaluates both task results and execution quality, including required tool coverage, dependency validity, and evidence traceability. Experimental results indicate that Urban-Agent reaches a 71% task success rate, 10 points above the strongest baseline. This lead holds across GPT-5-mini, Gemini-2.5-flash, DeepSeek-V4-flash, and Qwen3-235B-A22B.

arXiv

Forgetful Attention: An Auditable Support-Vector Memory for Selective Retention and Verified Deletion

Vishwajith Ramesh

Auditable memory requires a precise contract: which output is preserved, relative to which reference solve, and across which updates. We introduce Support Vector Attention (SV-Attention), a one-class support vector data description (SVDD) gate whose coefficients enter the readout. Zero-coefficient keys are reserve; positive-coefficient keys are active. Removing a reserve token without re-solving preserves the current readout. Maintained deletion, which updates the existing solver state, targets a fresh retained-key fit under the same coefficient cap C. Across 1,200 fp64 deletion/refit trials on Gaussian, redundant, MIMIC-IV, and learned keys, 1,199 complete. Median maximum gate-score discrepancy over declared probes ranges from 4.5e-13 to 5.7e-7, and fresh refitting is 24-223 times slower than maintained deletion on the reference CPU. A deterministic example shows that current reserve status does not guarantee equivalence after future admissions. At matched token counts, rare-group recall is 0.861 versus 0.319 for an oracle attention-mass proxy (H2O-style). In a held-out-channel ICU control, SpO2 below 90 percent defines events but SpO2 is excluded from every selector; event-hour retention is 0.464 versus 0.225 for an RBF-density baseline. A separate batched approximation supports end-to-end training; at 3.22M parameters, seven paired seeds yield mean best-validation bits per character of 2.178 versus 2.383 (p=0.001). The contracts are point-in-time and fixed-C; future-safe streaming and general-purpose performance remain open, and larger fixed-step runs suggest slower optimization.

arXiv

Leveraging Interpretable Tsetlin Machine for PDF Malware Detection

Rahul Jaiswal, Ole-Christoffer Granmo

In the digital era, Portable Document Format (PDF) is one of the most widely used file formats for storing and exchanging digital documents due to its platform independence and rich functionality. However, these same capabilities have also made PDF files an attractive attack vector for cyberattackers, who embed malicious code within seemingly legitimate documents to compromise target systems. This paper presents a novel interpretable Tsetlin Machine (TM)-based framework for PDF malware detection. The proposed framework extracts salient features from PDF documents through static analysis without executing the files and employs rule-based learning to accurately classify benign and malicious PDF documents. Numerical evaluation on the RIT-PDFMal-2026 dataset demonstrates that the proposed framework achieves an accuracy of 98.02%, outperforming several state-of-the-art machine learning classifiers. Moreover, the proposed framework provides intrinsic interpretability by transparently explaining its classification decisions. Edge deployment on a Raspberry Pi further supports real-time, on-device PDF malware detection. The combination of better accuracy, computational efficiency, and intrinsic interpretability makes the proposed framework a promising solution for practical PDF malware detection.

arXiv

ProPRL: Property-Aware Prerequisite Relation Learning in Educational Knowledge Graphs

Xinghe Cheng, Jiapu Wang, Chaobo He, Ruihai Dong, Quanlong Guan

Prerequisite relation learning is central to adaptive instruction, yet existing methods often formulate it as conventional link prediction, limiting their ability to adaptively integrate complementary educational evidence for individual candidate pairs and to discourage contradictory reverse predictions. We propose ProPRL, a Property-aware Prerequisite Relation Learning framework. ProPRL first learns complementary concept representations from a concept-resource hypergraph and a directed learning-behavior graph, where direction-preserving personalized propagation aggregates multi-hop behavioral evidence. It then employs a Pair-conditioned Gate to adaptively weight and fuse the two views for each candidate ordered concept pair. Finally, an \textit{Irreversibility Constraint} introduces an anti-symmetry regularizer that penalizes simultaneously high confidence in both directions of the same concept pair. Experiments on multiple real-world educational datasets show that ProPRL achieves state-of-the-art performance on prerequisite relation learning.

arXiv

On the missing benchmarks layer and a potential solution

Francis F Daniel, Mauro Ibañez, Francis Perelman, Marian Basti

Latin America is missing a foundational layer for native AI development: the benchmark layer. The benchmark layer does two things no other layer can - it audits AI systems against regional social requirements and it directs AI optimization in economically relevant environments. Without it, public institutions cannot independently evaluate foreign AI systems, and companies cannot optimize AI systems to solve local problems with SOTA performance. The cost of the missing layer is dual: a loss of auditability and a loss of optimization direction over a technology that is increasingly critical infrastructure. We propose an EvalsHub, with LatamBoard as its first regional instance - an open, task-first benchmark infrastructure where universities, public institutions, professional communities, and companies can publish, execute, compare, and maintain evaluations across models, workflows, and agents. Built once, measured forever - re-run by institutions as new AI systems ship and by industry teams after every system change. Open by design and incentive-driven by construction.

arXiv

SparSEEty: Extracting Tokens from Sparsity-Exploiting LLM Serving Systems via Deterministic Side Channels

Yongwan Jo, Jinyoung Park, Euihyun Lee, Dokyung Song

Modern large language models (LLMs) exhibit activation sparsity, wherein only a subset of their neurons is activated for given input tokens. Researchers have leveraged this property to optimize LLM serving systems by omitting weight accesses and computations pertaining to inactive neurons. Unfortunately, however, such optimizations create input-dependent weight accesses, which can be leaked over side channels. We present SparSEEty, a new token extraction attack that exploits input-dependent neuron weight accesses introduced by sparsity-exploiting LLM serving systems. SparSEEty first constructs a neuron-activation oracle using neuron weight access side channels during LLM inference, and then inverts the activation traces to reconstruct the input tokens, forming an end-to-end token extraction attack. We instantiate SparSEEty against an LLM serving system protected inside an Intel TDX confidential virtual machine (CVM), addressing three key challenges: (i) constructing a neuron-activation oracle using a combination of side channels exposed by CVMs, (ii) reducing inference-time overheads of neuron activation monitoring for covertness, and (iii) accurately inverting partial binary activation traces back to tokens. Our evaluation shows that SparSEEty can reconstruct both prompt and response tokens with consistently high BLEU scores (>0.95) across various models and datasets, while incurring monitoring overheads of 3.7% to 7.2%.

arXiv

Neurosymbolic Reasoning with Incremental Knowledge for Sample Efficient Hierarchical Reinforcement Learning

Subrat Prasad Panda, Blaise Genest, Arvind Easwaran

(Flat) Reinforcement Learning (RL) agents face significant challenges in environments with sparse rewards that require long-horizon reasoning. A compelling approach to improve sample efficiency is to incorporate knowledge into learning and decision-making. In standard Hierarchical RL (HRL), knowledge is encoded in a fixed, non-updatable form, such as architectural choices, and remains unchanged throughout learning. With fixed HRL, reasoning with incremental knowledge learned during exploration is impractical before sufficient environmental knowledge is acquired, leading to poor sample efficiency. In this work, we propose neurosymbolic HRL with {\em Incremental Knowledge (InK)}: symbolic high-level components perform {\em symbolic planning} (e.g. using $D^*$) on an updatable representation of current InK, while low-level goal-conditioned neural modules learn motion primitives through experience using reward shaping. Experiments on navigation tasks demonstrate that incorporating InK substantially improves sample efficiency. Additionally, to perform {\em optimal} symbolic planning given {\em prior} knowledge about the world, we develop Belief World Tree Search. The code is available at https://github.com/CPS-research-group/ink_bwts.

arXiv

Internalising the Identity Primitive: Cryptographic Individuality for an Autonomous Agent on a Public Blockchain

Keisuke Suzuki

A software agent on a public blockchain accumulates authority and economic stakes, raising the engineering question of what makes it count as an individual. The paper's central contribution is a shift of trust root for the key-to-weights binding of agent identity: from hardware, operator, or wrapper trust to cryptographic assumptions enforced by a pinned implementation (liveness, key custody, oracle trust, and the underlying software stack remain external). We design and deploy on Solana devnet an agent whose neural-network weights are a deterministic function of its private key. The binding is committed in zero knowledge at genesis, re-checked against that commitment at every state transition, and signed by the agent into an on-chain history unforkable once finalized; in a PoC-tier extension, a protocol-imposed metabolic cost is debited each cycle from a key-derived economic account, adding a consumption-side economic-viability constraint to the key-history-economy triple. Empirically, the agent completes a 2.36-day on-chain run with two host-side resumptions but no rejected transition, at bounded per-transition verification cost; a substituted substrate is rejected on chain, and independently keyed agents diverge as predicted while a same-key control stays at zero. To our knowledge, this is the first published on-chain agent whose identity primitive is itself a cryptographic invariant re-checked at every state transition. The resulting transition-time invariant instantiates the cryptographic individuality proposed by Suzuki 2026's Artificial Externality framework.

arXiv

HyperFL: Query-Adaptive Representation Learning for Software Fault Localization

Shuai Shao, Yiming Zeng, Yu Zhao, Tingting Yu

Software fault localization identifies the code locations responsible for reported issues and is a fundamental step toward automated debugging and program repair. Recent retrieval-based approaches formulate fault localization as a dense retrieval task by learning a shared embedding space between issue reports and source code. However, these methods encode all issue reports using a fixed query representation, despite the substantial diversity of real-world issue reports in length, structure, and debugging information. To address this limitation, we propose HyperFL, a query-adaptive representation learning framework for software fault localization. HyperFL employs a lightweight hypernetwork to generate query-specific LoRA parameters for the query encoder, enabling dynamic query adaptation while keeping the code encoder fixed and reusable. Experiments on a real-world issue localization benchmark demonstrate that HyperFL consistently improves retrieval performance across multiple embedding backbones, achieving up to 13.3% relative improvement in function-level MRR@10 and 16.7% relative improvement in Hit@1 over the state-of-the-art method SweRank. Further analysis shows that HyperFL learns distinct adaptation patterns for different issue characteristics, highlighting the effectiveness of query-adaptive representations for software issue localization.

arXiv

Trivium: Temporal Regret as a First-Class Objective for Causal-Memory Controllers

Edward Y Chang

Many agentic systems and LLM pipelines correct mistakes by optimizing outcome reward. This addresses only the what of failure; the why and when may go unlogged, allowing the same error to recur across episodes. We propose long-horizon temporal regret alongside outcome regret and epistemic regret. These are diagnostic quantities, not standard comparator-based online-learning regrets. Temporal regret captures how long an unresolved or incorrect causal model is tolerated; epistemic regret captures posterior error over that model. Over a stream of E episodes, we prove three conditional results under explicit probing, persistence, and detectability assumptions. First, under observationally equivalent confounding, outcome-only learning cannot separate causal from spurious structure, so miscalibration can persist after outcome regret reaches zero. Second, with a persistent causal log and budgeted probes, total probe complexity is logarithmic in E, inducing O(log E) delayed-identification temporal regret. The implemented clipped soft score adds a linear numerical floor, so this logarithmic claim applies only to identification delay. Third, under K detectable change-points, the rate extends to O(K log E). We instantiate Trivium and state five falsifiable predictions. On CausalBench-Seq, a hard structural readout records 7.8+-2.9 misidentified episodes per seed over 500 episodes and 20 seeds, with zero observed stationary errors, while outcome-only controllers remain misidentified throughout. Audit ablations show that continued posterior updating, not detector reopening or local repair, drives posterior recovery; local repair instead reduces committed-graph dispatch exposure. The prior logarithmic-envelope verdict based on a clipped soft score is withdrawn. A pilot real-LLM stream provides external evidence. Self-learning here means revising an external causal model, not retraining LLM weights.

arXiv

The Hamilton-Jacobi Theory of Deep Learning

Jose Marie Antonio Miñoza, Erika Fille T. Legara, Christopher P. Monterola

In this paper, training a neural network is identified, exactly, as a search through Hamilton--Jacobi initial-value problems: each gradient step selects the initial data of a viscous Hamilton--Jacobi equation whose Hopf--Cole propagator best fits the observations; at inference, the input is the spatial point at which that solution is evaluated and the initial condition is already encoded in the weights. The correspondence is exact for log-sum-exp layers, with ReLU, sigmoid, SiLU, and GELU each an exact limit, gradient, or moment of the same object, and exact in composition across depth and width, with a quantified error at finite depth that vanishes in the joint limit. It is structural for residual networks, transformers, and recurrent networks (RNNs, LSTMs, SSMs), each discretizing the same class of equations, at a named and quantified approximation error. A single deformation parameter $\varepsilon$ unifies all four perspectives (network, tropical algebra, viscous PDE, convex optimization) in a commutative diagram closed under Lipschitz conditions. Quantitative consequences include: the minimax optimal generalization rate $O(n^{-1/(d+2)})$ for fixed $t$; adversarial robustness controlled by $\varepsilon$; backpropagation as the co-state equation of the Hamiltonian system for residual networks (Pontryagin Maximum Principle); scaling exponents consistent with data intrinsic dimension via PDE quadrature; and a closed-form $O(N)$ influence function (softmax attribution weights $π_j$) whose entropy landscape undergoes fold bifurcations as $\varepsilon$ increases, each merging attribution basins.

arXiv

Tree of Thoughts as a Classical Heuristic Search Problem: Formal Foundations and Design Patterns

Guni Sharon

Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities, yet their standard generation process -- auto-regressive token prediction -- is inherently myopic and prone to cascading errors. To address this, the Tree-of-Thoughts (ToT) framework creates a search space over intermediate reasoning steps, allowing search models to explore, look ahead, and backtrack. However, current ToT research remains fragmented across Natural Language Processing and Automated Planning communities, often using inconsistent terminology and ad-hoc implementations. Consequently, we synthesize the ToT landscape through a unified taxonomy based on classical heuristic search terminology. We map LLM-based reasoning to classical search components: state representation (granularity of thoughts), successor generation (prompting operators), and heuristic evaluation (self-assessment of progress). We analyze existing work within the context of our taxonomy and identify emerging design patterns: systematic search (Best-First Search) for shallow, deterministic tasks and lookahead-heavy strategies (DFS, MCTS) for deep multi-step reasoning. We conclude by identifying open algorithmic challenges at the intersection of heuristic search and LLM reasoning, and call on the heuristic search community to engage with this emerging domain.

arXiv

A Comprehensive Evaluation of Code Language Models for Security Patch Detection

Nils Loose, Joseph Bienhüls, Kristoffer Hempel, Felix Mächtle, Thomas Eisenbarth

Automated detection of vulnerability-fixing commits (\vfcs) is critical for timely security patch deployment, as advisory databases lag patch releases by a median of 25 days and many fixes never receive advisories. Code language models are increasingly adopted for identifying VFCs, yet whether they can recognize a security fix from the code change itself remains unclear, as reported performance is shaped by commit messages, project-level data leakage, and uncertain data quality. We present a rigorous re-evaluation that jointly controls these factors through a unified framework consolidating 20 fragmented datasets spanning more than 180000 commits. Training 270 models from 125M to 80B parameters, we isolate the code signal under group-stratified code-only evaluation and assess the impact of model capacity and additional code context, ranging from intra-procedural enrichment to inter-procedural repository context. Model capacity yields clear but insufficient gains, and the evaluated context signals provide no reliable improvement under strict false positive budgets. At a false positive rate of 0.5%, every evaluated fine-tuned code-only model misses at least 80% of vulnerability fixes. A manual expert audit further shows that label error concentrates in commits lacking CVE association and primarily distorts evaluation. We derive concrete recommendations for evaluating on aggregated VFC datasets and release our unified framework and evaluation suite.

arXiv

Wedge Sampling: Efficient Tensor Completion with Nearly-Linear Sample Complexity

Hengrui Luo, Anna Ma, Ludovic STEPHAN, Yizhe Zhu

We introduce Wedge Sampling, a new non-adaptive sampling scheme for low-rank tensor completion. We study recovery of an order-$k$ low-rank tensor of dimension $n\times\cdots\times n$ from structured observations of its entries. Unlike the standard uniform entry model (i.e., i.i.d. samples from $[n]^k$), wedge sampling allocates observations to structured length-two patterns (wedges) in an associated bipartite sampling graph. By directly promoting these length-two connections, the sampling design strengthens the spectral signal that underlies efficient initialization, in regimes where uniform sampling is too sparse to generate enough informative correlations. Our main result shows that this change in sampling paradigm enables polynomial-time algorithms to achieve both weak and exact recovery with nearly linear sample complexity in $n$. The approach is also plug-and-play: wedge-sampling-based spectral initialization can be combined with existing refinement procedures (e.g., spectral or gradient-based methods) using only an additional $\tilde O(n)$ uniformly sampled entries, substantially improving over the $\tilde O(n^{k/2})$ sample complexity typically required under uniform entry sampling for efficient methods. We also formulate a noisy wedge-sampling extension for additive Gaussian observations and analyze both the spectral and gradient-descent procedures under suitable signal-to-noise conditions. Thus, the computational barrier in tensor completion is sensitive to the observation model: while it persists under uniform entry sampling, it can be bypassed by non-adaptive structured designs that provide a stronger initialization.

arXiv

Explainability Methods for Hardware Trojan Detection: A Systematic Comparison

Paul Whitten, Francis Wolff, Chris Papachristou

Hardware trojans are malicious circuits which compromise the functionality and security of an integrated circuit (IC). These circuits are manufactured directly into the silicon and cannot be fixed by security patches like software. The solution would require a costly product recall by replacing the IC and hence, early detection in the design process is essential. Hardware detection at best provides statistically based solutions with many false positives and false negatives. These detection methods require more thorough explainable analysis to filter out false indicators. Existing explainability methods developed for general domains like image classification do not always provide the actionable insights hardware engineers need. A question remains: how do domain-aware property analysis, model-agnostic case-based reasoning, and model-agnostic feature attribution techniques compare for hardware security applications? This work compares three categories of explainability for gate-level hardware trojan detection on the Trust-Hub benchmark dataset: (1) domain-aware property-based analysis of 31 circuit-specific features derived from gate fanin patterns, flip-flop distances, and primary Input/Output (I/O) connectivity; (2) model-agnostic case-based reasoning using k-nearest neighbors for precedent-based explanations; and (3) model-agnostic feature attribution methods (Local Interpretable Model-agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), gradient) that provide generic importance scores without circuit-level context.

arXiv

PSI3D: Plug-and-Play 3D Stochastic Inference with Slice-wise Latent Diffusion Prior

Wenhan Guo, Jinglun Yu, Yaning Wang, Jin U. Kang, Yu Sun

Diffusion models are highly expressive image priors for Bayesian inverse problems. However, most diffusion models cannot operate on large-scale, high-dimensional data due to high training and inference costs. In this work, we introduce a Plug-and-play algorithm for 3D stochastic inference with latent diffusion prior (PSI3D) to address massive ($1024\times 1024\times 128$) volumes. Specifically, we formulate a Markov chain Monte Carlo approach to reconstruct each two-dimensional (2D) slice by sampling from a 2D latent diffusion model. To enhance inter-slice consistency, we also incorporate total variation (TV) regularization stochastically along the concatenation axis. We evaluate our performance on optical coherence tomography (OCT) super-resolution. Our method significantly improves reconstruction quality for large-scale scientific imaging compared to traditional and learning-based baselines, while providing robust and credible reconstructions.

arXiv

SkillTrace: Traversing a Query-Skill Graph for Composable LLM Agents

Yue Yao, Shengyuan Wang, Xin Chen, Minke Zhang, Jia He, Bingjun Luo, Tom Gedeon

Large language model agents increasingly solve complex tasks by composing reusable skills from a library. To address this, the key challenge is not merely to retrieve individually relevant skills, but to identify a complete and executable skill composition. In this paper, we argue that this problem can be solved in a graph with three levels: compositional relations among skill queries, similarity between queries and candidates in the skill library, and the dependencies among the selected candidates. We introduce SkillTrace, which organizes the user query into a semantic hierarchy, matches skill queries and candidates, and propagates over the skill dependencies. Experiments on SkillsBench and ALFWorld demonstrate that SkillTrace achieves state-of-the-art performance, reaching a success rate of 53.17% on SkillsBench and 91.43% on ALFWorld. SkillTrace also delivers consistent improvements across different backbone language models, demonstrating the generality and robustness of graph-based skill retrieval.

arXiv

Before Reasoning Can Fail: Pre-Evidence Procedural Failures in Agentic RAG

Daeyoung Roh, Donghee Han

Agentic retrieval-augmented generation (RAG) systems can fail before evidence-conditioned reasoning is tested: an agent may retrieve candidate snippets but finalize without inspecting them. We study this failure mode as a procedural property of the agent trajectory, decomposing wrong answers into pre-evidence discipline failures and post-gold-read failures using saved tool-call traces, retrieved evidence, read passages, and final answers. Across 12,000 paired trajectories on HotpotQA, 2WikiMultiHopQA, and MuSiQue, the two failure types are largely non-redundant: the both-trigger rate is in [11.2%, 13.1%] across regex and spaCy entity extractors. We then evaluate Read-Gate, a minimal runtime invariant requiring an agent to read after search and before finalization. Forced reading improves LLM-Acc by 14.9-19.9 points on trajectories that would otherwise skip reading and by 3.2-9.4 points on full minimal-reasoning cells. Additional diagnostics show that larger hidden thinking budgets do not necessarily increase evidence inspection. Together, these results indicate that evidence-gathering should be evaluated as a trajectory-level control problem, separately from answer-side reasoning.

arXiv

Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMs

Zixuan Huang, Yang Zhou, Kaixuan Wang, Guli Zhang, Hongyan Xie, Yakun Zhu, Hao GENG, Xiaozhi Chen, Yikun Ban, Deqing Wang

Recent Vision-Language-Action (VLA) models for autonomous driving (AD) increasingly utilize chain-of-thought (CoT) supervision to enhance the reasoning capabilities of their Vision-Language Model (VLM) components, yet existing annotation pipelines commonly expose the teacher model to the logged ground-truth (GT) future trajectory. We empirically show that this induces trajectory anchoring bias: teacher models rationalize the revealed outcome rather than infer a decision from scene evidence, producing less causally faithful CoTs and substantially more severe hallucinations, especially in causally challenging scenes. Removing the GT trajectory eliminates this shortcut, but open-ended trajectory generation entangles high-level decision-making with precise geometric synthesis and low-level dynamics. To make trajectory-level driving decisions verifiable without requiring open-ended trajectory synthesis, we introduce Autonomous-Driving Multiple-Choice Question (AD-MCQ), which casts planning as selection among explicit trajectory candidates. Taking this a step further, we propose Deferred Exposure of Future Trajectories for RLVR (DEFT-RLVR) to transform future trajectories from pre-decision anchors into post-decision verification targets. Experimental results show that DEFT-RLVR improves AD reasoning while preserving or even enhancing general visual capabilities. With VLM-only inference and controllable difficulty through candidate construction, AD-MCQ provides a flexible, scalable, and extensible foundation for future research on verifiable AD reasoning.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.