Author Correction: MAGE-A4/MAGE-A8-targeted TCR-based bispecific T cell engager in recurrent and/or refractory solid tumors: a phase 1 trial
该条目无摘要(多为勘误、社论或新闻类内容,出版方未提供摘要)
Academic Intelligence · Curated Daily
AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。
该条目无摘要(多为勘误、社论或新闻类内容,出版方未提供摘要)
arXiv:2406.11783v2 Announce Type: replace-cross Abstract: We study the systole of a model of random hyperbolic 3-manifolds introduced by Petri and Raimbault, answering a question posed in that same article. These are compact manifolds with boundary constructed by randomly gluing truncated tetrahedra along their faces. We prove that the limit, as the volume tends to infinity, of the expected value of their systole exists and we give a closed formula of it. Moreover, we compute a numerical approximation of this value.
arXiv:2606.12410v1 Announce Type: cross Abstract: We study the random variable that counts the number of specific arrangements of clustered consecutive numbers in permutations under the Mallows distribution. We provide an asymptotic expression for the expected value of this random variable. This result extends and tightens the previously known result by Pinsky (2022) concerning clustered consecutive numbers in Mallows permutations. Moreover, we identify a range of parameters for which the distribution of the number of arrangements of clustered consecutive numbers in Mallows permutations is close to a Poisson distribution.
Background Upper airway disease is common in primary ciliary dyskinesia (PCD), but management evidence is limited. We aimed to describe management practices and identify factors influencing management decisions. Methods Using data from the Ear-Nose-Throat (ENT) Prospective International Cohort of patients with PCD (EPIC-PCD) and an ENT-specialist survey across participating centres, we described management practices recorded at routine follow-up. We assessed clinical factors associated with practices via mixed-effects logistic regression models. In a subgroup of patients, we assessed factors associated with initiation or discontinuation of practices. Results We included 579 patients: median age 15 years, 46% female. Nasal rinsing (54%) and nasal corticosteroids (22%) were most frequently prescribed. Among 466 patients with available data, 47 had grommets (10%) and 42 hearing aids (9%). Nasal corticosteroids and rinsing were more frequently prescribed in patients with polyps (odds ratio [OR] 3.74, 95% confidence interval [CI] 1.80-7.76; OR 3.39, 95% CI 1.37-8.37) or turbinate hypertrophy (OR 1.89, 95% CI 1.03-3.47; OR 2.89, 95% CI 1.55-5.38), and upper airway nebulisation in patients with frequent nasal symptoms (OR 2.86, 95% CI 1.11-7.39). Management practices differed between centres, as seen also by the specialists survey responses. In 177 patients with multiple visits, initiation of nasal rinsing was associated with frequent nasal symptoms (OR 3.18, 95% CI 1.24-8.18) and turbinate hypertrophy (OR 3.21, 95% CI 1.20-8.59). Conclusion Upper airway disease management in PCD varies and is partly guided by symptom burden and clinical findings. This variation across centres highlights the need for care standardisation and PCD-specific management guidelines.
Despite advances in object detection, aerial imagery remains a challenging domain, as models often fail to generalize across variations in spatial resolution, scene composition, and semantic label coverage. Differences in geographic context, sensor characteristics, and object distributions across datasets limit the capacity of conventional models to learn consistent and transferable representations. Shared methods trained on such data tend to impose a unified representation across fundamentally different domains, resulting in poor performance on region-specific content and less flexibility when dealing with novel object categories. To address this, we propose a novel modular learning framework that enables structured specialization in aerial detection. Our method introduces a hierarchical routing mechanism with two levels of modularity: a domain routing layer that uses latent geographic embeddings to assign inputs to domain-specialized expert modules, and a scene routing mechanism that allocates image subregions to scene-specific expert modules. This allows our method to specialize across datasets and within complex scenes. Additionally, the framework contains a conditional expert module that uses external semantic information (e.g., category names or textual descriptions) to enable detection of novel object categories during inference, without the need for retraining or fine-tuning. By moving beyond monolithic representations, our method provides an adaptive framework for remote sensing object detection. Comprehensive evaluations on four datasets highlight improvements in multi-dataset generalization, region-level specialization, and open-category detection.
In conventional hearing-aid personalisation, clinicians cannot hear what their patients hear, and patients cannot often reliably detect or describe what they hear. Self-adjustment avoids this issue but requires user controls that adjust hearing-aid signal processing parameters to be effective, efficient and easy. In this study, we explored (a) the roles of interface complexity and stimulus type in the self-adjustment of hearing-aid gain, and (b) how well individuals can adjust one sound to match another to assess the same interfaces and stimuli. Adult hearing-aid users with mild to moderate symmetrical sensorineural hearing loss repeatedly adjusted the gain (a) to their preference from individual prescription (n = 41) and (b) to match their previous preferences from a random starting point (n = 32) using three interfaces representing different bass/mid/treble configurations and three stimuli (music, speech and speech-in-noise). The large interindividual variability in self-adjusted gains clustered into three patterns of deviation from initial prescription: increased relative bass, overall gain reduction, and close to initial prescription. There were no substantial effects of interface nor stimulus on self-adjustment reliability (median {sigma} = 2.8 dB), whereas absolute sound-matching error increased with increasing interface complexity and centre frequency. Neither individual matching accuracy nor questionnaire responses predicted either self-adjusted gains or reliability. Overall, these results show that many - but not all - hearing-aid users can adjust gains with reasonable reliability, and while it can be difficult to predict the behaviour from the individual, the individual applies a similar self-adjustment behaviour across different interfaces and stimuli.
arXiv:2606.20532v1 Announce Type: new Abstract: Style-captioned text-to-speech systems use natural language to control voice characteristics, but how individual words influence acoustic output remains unclear. Understanding this is critical for diagnosing failure modes and improving controllability in expressive TTS. We propose cross-attention attribution for speech diffusion models, adapting the DAAM framework to the speech domain for the first time, and apply it to CapSpeech-TTS. Our method extracts per-token heatmaps across 25 layers and 24 ODE steps. We analyze 3,600 (style caption, text transcript) combinations comprising 120 style captions conditioning the generation of 30 text transcripts each, revealing how caption tokens shape waveforms. Results show: (1) style tokens have lower temporal variance than content/function tokens, confirming global conditioning; (2) style attention correlates with F0 and energy; (3) style conditioning peaks in early steps and deep layers; (4) attention entropy reaches its minimum at layer 17, co-occurring with the style importance peak, indicating maximal network selectivity at the most style-critical stage. This is the first study of how natural language influences cross-attention in speech diffusion models
arXiv:2606.04009v2 Announce Type: replace-cross Abstract: Two-sample testing is a fundamental tool for detecting distributional differences across scientific domains, but classical tests (including kernel-based tests) can be ineffective on high-dimensional structured data such as images. Recent deep two-sample tests improve sensitivity in these settings by learning informative representations, yet they provide limited insight into which data features drive rejection of the null hypothesis $H_0$. To address this issue, we propose a counterfactual explanation framework for deep two-sample testing that generates sample-level edits moving observations from a source group toward a target group while explicitly reducing the discrepancy measured by the test. Our method combines a diffusion autoencoder with a pretrained deep two-sample test model and optimizes a maximum mean discrepancy (MMD) objective in the test model's representation space to produce plausible counterfactuals. We quantify distribution-level effects through changes in the test statistic and the resulting two-sample p-values. We evaluate the method on synthetic 2D shape datasets and two MRI cohorts. Across both settings, the counterfactual transformations consistently increase p-values relative to the original samples, indicating that the edited source set becomes statistically closer to the target distribution under the test. We measure minimality using LPIPS to ensure the counterfactuals remain close to the original samples. The resulting edits provide interpretable evidence of the features associated with the detected group differences. On MRI, the localized changes are consistent with known anatomical differences between cohorts.
arXiv:2606.11379v1 Announce Type: new Abstract: Pre-mediation, the preparatory phase preceding direct human negotiation, plays a critical role in achieving mutually beneficial agreements, yet is often omitted due to cost, time, and limited access to trained mediators. We introduce an automated mediator for human negotiation, implemented as a structured pipeline of LLM modules, that supports pre-mediation in integrative negotiation settings. The pipeline decomposes preparation into specialized modules for dialogue, preference prediction, response-level critique, and structured summarization, separating inference, generation, and evaluation to address limitations of monolithic single-prompt approaches. We use the term "agent" for each module following common LLM-systems terminology, but the components are not autonomous and do not interact peer-to-peer; outputs are passed forward in a fixed sequence. We evaluate the system in two controlled human-subject experiments comparing AI-based pre-mediation with professional human mediators in a multi-issue negotiation scenario. On short-term self-reported measures, the automated mediator achieves preparation outcomes broadly comparable to human mediators, including trust in the mediator and confidence in reaching mutually beneficial agreements, while achieving substantially lower error on the preference-inference task under our scenario and prompts (36% lower RMSE). A second study shows that targeted prompt refinements reduce excessive affirmation patterns from 36.6% to 16.8%, matching human mediator baselines. Our findings suggest that structured LLM pipelines can provide scalable, low-effort pre-mediation support broadly comparable to human mediators on short-term self-reported preparation outcomes. The pipeline's single-party design mirrors how human mediators run pre-mediation today and enables parallel deployment across all parties to a dispute, supporting scalability.
arXiv:2606.11793v1 Announce Type: cross Abstract: Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, partly driven by the uncertainties affecting the land surface representation and variability in Earth system models. To address this limitation, we present a data-driven framework AI4Land, for generating high-resolution historical reconstructions and future projections of key land surface variables. The framework follows a two-phase approach using a U-Net architecture. In the first phase, which is the focus of this work, it reconstructs annual land use and land cover by integrating coarse-resolution scenario data with static geophysical features. In a planned second phase, the resulting high-resolution maps will be used to predict dynamic biophysical variables, particularly leaf area index, at finer temporal scales. Trained on Earth observation data, the models learn to reproduce spatially explicit and physically consistent land surface patterns, extending temporal coverage to periods lacking direct observations. AI4Land was developed and trained on MareNostrum5, demonstrating how GPU-accelerated HPC infrastructure enables global-scale climate AI pipelines. The final product is a suite of open-source emulators designed for real-time coupling with digital twin platforms, such as those developed under the Destination Earth initiative. By delivering realistic and evolving land surface conditions on demand, this work aims to reduce critical uncertainties and improve the predictive power of next-generation climate simulations.
arXiv:2603.28387v2 Announce Type: replace Abstract: Trustworthy clinical AI requires that performance gains reflect genuine evidence integration rather than surface-level artifacts. We evaluate 12 open-weight vision-language models (VLMs) on binary classification across two clinical neuroimaging cohorts, \textsc{FOR2107} (affective disorders) and \textsc{OASIS-3} (cognitive decline). Both datasets come with structural MRI data that carries no reliable individual-level diagnostic signal. Under these conditions, smaller VLMs exhibit gains of up to 58\% F1 upon introduction of neuroimaging context, with distilled models becoming competitive with counterparts an order of magnitude larger. A contrastive confidence analysis reveals that merely mentioning MRI availability in the task prompt accounts for 70-80\% of this shift, independent of whether imaging data is present, a domain-specific instance of modality collapse we term the scaffold effect. Expert evaluation reveals fabrication of neuroimaging-grounded justifications across all conditions, and preference alignment, while eliminating MRI-referencing behavior, collapses both conditions toward random baseline. Our findings demonstrate that surface evaluations are inadequate indicators of multimodal reasoning, with direct implications for the deployment of VLMs in clinical settings.
arXiv:2606.16662v1 Announce Type: new Abstract: We prove an abstract operator-level convergence criterion for mixed paracontrolled random operators \[ T^{i;j,k}_{\Lambda}(w) = I_i(w
Clonal haematopoiesis (CH) activates inflammation and increases the risk of atherosclerosis1,2. Whether lifestyle alters CH clone expansion or the phenotypic programming of CH mutant cells, thereby affecting atherosclerosis, is unknown. Here, in humans and mice and across mutations in Jak2, Tet2, Trp53 and Dnmt3a, we demonstrate mutation-dependent responses to sleep and exercise in CH and show that mutant cells are uniquely sensitive to lifestyle. In two human datasets, moderate-to-vigorous physical activity was associated with lower prevalence of non-DNMT3A-driven CH. In atherogenic mice with Jak2V617F or Tet2 loss of function (LOF), but not Trp53 LOF or Dnmt3aR878H CH, uninterrupted sleep or exercise curtails clone expansion. In CH with the Jak2V617F mutation, sleep and exercise reduces clone expansion by selectively reprogramming mutant, but not cohabitant wild type, haematopoietic progenitor cells towards antiproliferative and metabolically healthy phenotypes by tempering bone marrow macrophage–haematopoietic progenitor cell IL-1β signalling. Sleep or exercise also lessens Jak2V617F-driven, Tet2 LOF-driven and Trp53 LOF-driven, but not Dnmt3aR878H-driven, atherosclerosis by locally reprogramming mutant vascular macrophages, independent of peripheral clone dynamics. In Jak2V617F, but not adjacent wild type, aortic macrophages, uninterrupted sleep blunts CLEC4E-dependent inflammasome activation, consequently diminishing lesions. Exercise, meanwhile, activates PAC1+ neurons in the locus coeruleus, raising the levels of peripheral noradrenaline, which signals through adrenergic receptor β2 (ADRβ2) whose expression is preserved by exercise in Jak2V617F, but not cohabitant wild type, aortic macrophages, selectively repressing their inflammatory programming and atherosclerosis. Our findings establish that healthy lifestyles gene-specifically diminish CH and selectively reprogram mutant haematopoietic progenitor cells and macrophages to maintain cardiovascular health. Sleep and exercise can slow clonal haematopoiesis and limit mutant cell-driven atherosclerosis.
Generative AI has enabled the creation of photorealistic images and videos that are increasingly disseminated on social media, often used for spam, misinformation, manipulation, and fraud. Existing AI-generated content (AIGC) detection methods face challenges including poor generalization to new generation models, reliance on single modalities, and lack of interpretable explanations. We present our pipeline that mitigates these issues by continuously curating diverse multi-modal social media data and training a compact vision-language model for detection and explanation. Our model achieves state-of-the-art detection performance on public benchmarks and demonstrates robust detection and explanation capabilities on internal social media datasets across multiple platforms. We deployed our model for post recommendation on social media platforms and observed positive downstream impacts on user engagement, demonstrating that it is feasible to perform effective AIGC detection in dynamic, real-world social media environments.
arXiv:2606.12207v1 Announce Type: cross Abstract: Embodied intelligence now spans navigation, household assistance, manipulation, autonomous driving, aerial agents, and multimodal large-model control. This expansion has made benchmark construction a central bottleneck for reliable evaluation. Unlike static datasets, embodied benchmarks combine task specifications, environments, robot data, demonstrations, annotations, metrics, evaluation scripts, and release policies into a single evaluation system. This survey reviews the literature through a five-stage construction pipeline: requirement and task construction, data acquisition, data cleaning and annotation, benchmark suite generation and metric definition, and evaluation execution with diagnostic feedback. For each stage, the survey analyzes the transition from manual curation to traditional automation, foundation-model assistance, and agentic closed-loop workflows. It also compares qualitative construction costs across human labor, data and asset acquisition, compute and simulation, validation and debugging, governance and maintenance, and rework risk. The main conclusion is that automation does not simply reduce benchmark cost. Instead, it often shifts cost toward validation, auditability, version control, and long-term governance. Progress in embodied evaluation will therefore depend not only on larger benchmark suites, but also on construction pipelines that are diagnosable, auditable, and responsibly refreshable.
arXiv:2604.02751v2 Announce Type: replace Abstract: Diffusion models often degrade in latent spaces, yet the formal causes remain poorly understood. We quantify latent-space diffusability via the rate of change of the Minimum Mean Squared Error (MMSE) along the diffusion trajectory. Our framework decomposes this MMSE rate into contributions from Fisher Information (FI) and Fisher Information Rate (FIR). We demonstrate that while global isometry ensures FI alignment, FIR is governed by the interplay between encoder and data geometries. Our analysis decouples diffusion degradation into four penalties: dimensional compression, tangential distortion, high-frequency encoder curvature, and intrinsic data curvature. We derive theoretical conditions for FIR preservation to ensure stable diffusability. Experiments across diverse autoencoding architectures demonstrate the implications of our theoretical bounds. We establish FI and FIR as a comprehensive analytical framework for understanding latent diffusability.
Introduction: Adverse drug withdrawal events (ADWEs) are a key safety concern with deprescribing but are infrequently reported in trials. Although pharmacovigilance systems have advanced our understanding of medication-related harms, it is unclear how extensively these systems have been used for ADWEs. Objectives: To examine the reporting patterns of ADWEs for all drugs recorded in United States and European pharmacovigilance databases between 2004 and 2023. Methods: A retrospective study was conducted using two pharmacovigilance databases, the publicly available FDA-FAERS dataset and EMA-EV Level 2A (individual-level) dataset. ADWE cases were identified using relevant MedDRA preferred terms. Data on patient characteristics, reporter type, drugs, indication, ADWE outcomes, dechallenge/rechallenge, seriousness criteria, time to onset, duration, and causality were summarised. Results: A total of 158,505 ADWE reports were analysed (FDA-FAERS: 145,514; EMA-EV: 12,987), with mean ages of 46.1 (FDA; 55.3% female) and 45.5 years (EMA; 57.1% female). The frequently reported drug classes were opioids (FDA: oxycodone, 29.8%; EMA: buprenorphine, 19%), antidepressants (FDA: duloxetine, 32%; EMA: venlafaxine, 25.9%) and gabapentinoids (FDA: pregabalin, 6.7%; EMA: pregabalin, 6.0%). The most common adverse outcomes were other serious medical conditions (FDA=63.9%; EMA=46.0%), hospitalisation (FDA=15.9%; EMA=28.3%), and disability (FDA=13.3%; EMA=6.2%) and these outcomes varied significantly based on sex and age group (p
Reinforcement learning (RL) has become a key component in modern large language models, yet the rollout stage remains the key bottleneck in RL training pipelines. Although Multi-Token Prediction (MTP) offers a natural solution to accelerate rollouts through speculative decoding, many studies have observed that MTP acceptance rates degrade significantly during RL training, leading to limited speedup performance. To address this bottleneck, we present Bebop, a systematic study of MTP in LLM post-training, and offer practical recipes to integrate MTP into large-scale RL pipelines. First, we reveal that the MTP acceptance rate is fundamentally bounded by the fluctuation of model entropy, which demonstrates a clear negative linear relationship with the rise of entropy in the RL stage. Second, we show that probabilistic rejection sampling largely alleviates the disturbance introduced by entropy in RL compared to greedy draft sampling. We further identify that the conventional MTP training objectives (cross-entropy or KL) are suboptimal in such settings, and therefore we propose a novel end-to-end TV loss that directly optimizes multi-step rejection sampling acceptance rate, yielding ~10% acceptance rate improvements, achieving up to 95% acceptance rates and up to 25% extra inference throughput gains across mathematical reasoning, code generation, and agentic tasks. Third, we test various online MTP training strategies during RL and show that pre-RL MTP training with e2e TV loss and rejection sampling achieves a consistent acceptance rate and speedup throughout the entire RL, eliminating the need for costly online MTP updating. We provide extensive experiments and analysis that validate our findings. Experimental results show our method achieves up to 1.8x end-to-end acceleration in async RL training of Qwen3.5, Qwen3.6, and Qwen3.7 models.
arXiv:2604.13085v2 Announce Type: replace-cross Abstract: Autonomous AI agents operating in dynamic environments face a persistent challenge: acquiring new capabilities without erasing prior knowledge. We present Adaptive Memory Crystallization (AMC), a memory architecture for progressive experience consolidation in continual reinforcement learning. AMC is conceptually inspired by the qualitative structure of synaptic tagging and capture (STC) theory, the idea that memories transition through discrete stability phases, but makes no claim to model the underlying molecular or synaptic mechanisms. AMC models memory as a continuous crystallization process in which experiences migrate from plastic to stable states according to a multi-objective utility signal. The framework introduces a three-phase memory hierarchy (Liquid–Glass–Crystal) governed by an Itô stochastic differential equation (SDE) whose population-level behavior is captured by an explicit Fokker–Planck equation admitting a closed-form Beta stationary distribution. We provide proofs of: (i) well-posedness and global convergence of the crystallization SDE to a unique Beta stationary distribution; (ii) exponential convergence of individual crystallization states to their fixed points, with explicit rates and variance bounds; and (iii) end-to-end Q-learning error bounds and matching memory-capacity lower bounds that link SDE parameters directly to agent performance. Empirical evaluation on Meta-World MT50, Atari 20-game sequential learning, and MuJoCo continual locomotion consistently shows improvements in forward transfer (+34–43\% over the strongest baseline), reductions in catastrophic forgetting (67–80\%), and a 62\% decrease in memory footprint.
arXiv:2506.14202v4 Announce Type: replace-cross Abstract: End-to-end backpropagation requires storing activations throughout all layers, creating memory bottlenecks that limit model scalability. Existing block-wise training methods offer means to alleviate this problem, but they rely on ad-hoc local objectives and remain largely unexplored beyond classification tasks. We propose $DiffusionBlocks$, a principled framework for transforming transformer-based networks into genuinely independent trainable blocks that maintain competitive performance with end-to-end training. Our key insight leverages the fact that residual connections naturally correspond to updates in a dynamical system. With minimal modifications to this system, we can convert the updates to those of a denoising process, where each block can be learned independently by leveraging the score matching objective. This independence enables training with gradients for only one block at a time, thereby reducing memory requirements in proportion to the number of blocks. Our experiments on a range of transformer architectures (vision, diffusion, autoregressive, recurrent-depth, and masked diffusion) demonstrate that DiffusionBlocks training matches the performance of end-to-end training while enabling scalable block-wise training on practical tasks beyond small-scale classification. DiffusionBlocks provides a theoretically grounded approach that successfully scales to modern generative tasks across diverse architectures. Code is available at https://github.com/SakanaAI/DiffusionBlocks .
Human migration is a fundamental driver of global demographic change, shaping population structure, labour markets and social policy across countries1–3. Although long-term migration patterns are often linked to economic development4, they can shift rapidly in response to shocks such as conflict, environmental crises and political change5. Despite its importance, migration remains difficult to measure consistently: existing data are sparse, concentrated in high-income settings and are fragmented across incompatible definitions, temporal resolutions and data types6–8. Past efforts have relied on partial datasets, including flow records, stock estimates and model-based reconstructions with limited coverage9–14. A central challenge is therefore to construct a globally consistent, high-resolution account of migration flows over time. Here we present a new dataset of annual origin-destination migration across 230 countries and regions from 1990 to the present, integrating diverse data sources into a unified modelling framework. By combining official statistics, census-based stocks, net migration estimates and past flow reconstructions, our approach produces temporally detailed and spatially comprehensive estimates that substantially extend existing resources. Using an ensemble of deep recurrent neural networks informed by geographic, economic, cultural and political covariates, we capture both persistent trends and short-term responses to changing conditions—all while propagating uncertainty to generate confidence bounds. Our results outperform existing five-year flow estimates on held-out data and provide finer temporal resolution, revealing previously obscured dynamics in global migration patterns. This framework highlights regions in which uncertainty remains high and data collection is most urgently needed. By releasing all data, code and trained models, we provide a transparent and reproducible foundation for future work. These advances enable a more timely and detailed understanding of human mobility, with implications for research and policy in an increasingly dynamic global system. A global annual migration-flow dataset (1990–2024) is produced using deep-learning models and diverse sources to estimate movements across 230 countries with improved temporal resolution, coverage and uncertainty estimates.
arXiv:2508.05287v3 Announce Type: replace-cross Abstract: Existing time series foundation models (TSFMs), often based on transformer variants, lack adaptability to different sampling rates, struggle with generalization across varying context and target lengths, and are computationally inefficient. We introduce FlowState, a novel TSFM architecture that achieves sampling-rate-equivariant forecasting through a unified design that pairs a state space model (SSM) encoder with a functional basis decoder (FBD). This design enables continuous-time modeling and dynamic time-scale adjustment, allowing FlowState to inherently generalize across all possible temporal resolutions, and dynamically adjust the forecasting horizons without retraining. We further propose an efficient pretraining strategy that improves robustness and accelerates training. Despite being one of the smallest TSFMs, FlowState achieves state-of-the-art results on the widely used GIFT-Eval benchmark, while demonstrating superior adaptability to unseen sampling rates. Our detailed analyses confirm the effectiveness of its components, and we demonstrate its unique ability to adapt to varying input sampling rates.
arXiv:2606.10616v2 Announce Type: replace Abstract: Long-horizon language agents accumulate observations, reasoning traces, and retrieved facts that exceed their finite context windows, making memory retention a fundamental resource-allocation problem. Existing memory systems improve management through heuristic scoring, retrieval optimization, or learned compression, but largely treat retention as a local decision problem and do not explicitly model its long-term consequences under realistic observability constraints. To fill this gap, we formulate memory retention as a constrained stochastic optimization problem with explicit budget feasibility, evidence utility, and delayed costs including miss penalties, reacquisition delays, and stale-information risk. We then propose OSL-MR (Observability-Safe Learning for Memory Retention), a novel framework that enforces a strict separation between online-observable features and offline-available supervision (OAS). OSL-MR combines an evidence learner trained from realized evidence supervision with a Mixed-Score heuristic that serves both as a deployable online-safe baseline and as a structured inductive prior for learning. The resulting policy learns query-conditioned evidence value directly from interaction data while remaining deployable under the same observability constraints. Experiments on LOCOMO and LongMemEval show that OSL-MR consistently outperforms recency-based methods, Generative Agents-style scoring, and other heuristic baselines, particularly under tight memory budgets. The Mixed-Score prior further improves precision while preserving recall, and sensitivity analysis demonstrates robustness across a wide range of cost configurations.
arXiv:2606.17328v1 Announce Type: new Abstract: LLM agents increasingly maintain long-term memory of user facts across sessions. Yet such memory is usually evaluated by aggregating accuracy over question rows or episodes. Because this approach scores question rows independently, even when several questions probe the same fact, it cannot show how that fact behaves as conditions change. We introduce MemTrace, a benchmark whose unit of measurement is the knowledge point: a single typed fact about the user, rather than an individual question. MemTrace probes each fact along three controlled dimensions: memory age, defined by how many sessions ago the fact appeared in the history; question type, covering current state, earlier state, and trajectory of change; and evidence condition, covering present, missing, and contradicted-by-false-premise settings. Evaluating 13 memory-system configurations across four paradigms, we find that similar pooled accuracy hides different failures: recovering a fact's current and earlier states does not imply tracking how it changed, and safe abstention does not imply correcting a false premise. The dominant bottleneck is evidence use, not retrieval: when systems fail, the evidence was retrievable 10 times more often than it was missing. These results suggest that improving long-term memory requires better use of reachable evidence, not simply more storage or retrieval.
arXiv:2601.21714v5 Announce Type: replace Abstract: The evolution of Large Language Model (LLM) agents towards System~2 reasoning, characterized by deliberative, high-precision problem-solving, requires maintaining rigorous logical integrity over extended horizons. However, prevalent memory preprocessing paradigms suffer from destructive de-contextualization. By compressing complex sequential dependencies into pre-defined structures (e.g., embeddings or graphs), these methods sever the contextual integrity essential for deep reasoning. To address this, we propose E-mem, a framework shifting from Memory Preprocessing to Episodic Context Reconstruction. Inspired by biological engrams, E-mem employs a heterogeneous hierarchical architecture where multiple assistant agents maintain uncompressed memory contexts, while a central master agent orchestrates global planning. Unlike passive retrieval, our mechanism empowers assistants to locally reason within activated segments, extracting context-aware evidence before aggregation. Evaluations on the LoCoMo benchmark demonstrate that E-mem achieves over 54\% F1, surpassing the state-of-the-art GAM by 7.75\%, while reducing token cost by over 70\%.