Predictive Memory Localization: Forecasting Selective Intervention Paths from Internal Signals

• URL: http://arxiv.org/abs/2608.12892v2
• Date: Fri, 14 Aug 2026 12:31:30 GMT
• Title: Predictive Memory Localization: Forecasting Selective Intervention Paths from Internal Signals
• Abstract summary: We introduce Predictive Memory Localization (PML), which treats the measured-grid intervention path as the predictive object of memory localization.<n>PML separates random-calibrated target movement from semantic-neighbor and capability damage, and compares static localization and supervised geometry with a strength-disjoint low-dose causal response.<n>Across three residual-norm-matched base models, learned directions retain selective-path gains and low-dose responses yield 0.801-0.828 record-held-out macro AUROC.
• Score: 4.075831906880532
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Activation steering turns localized representations into control directions, but localization alone does not reveal whether a direction has a selective operating regime. We introduce Predictive Memory Localization (PML), which treats the measured-grid intervention path as the predictive object of memory localization. PML separates random-calibrated target movement from semantic-neighbor and capability damage, and compares static localization and supervised geometry with a strength-disjoint low-dose causal response. Our frozen study covers 3,000 records from nine datasets and fourteen domains, yielding 30,000 distinct record-direction-layer paths and 210,000 distinct path-strength evaluations. At layer 7, the geometry-derived RFM/AGOP direction reaches 13.1% target-any and 12.3% clean-any, exceeding random by 3.6 and 3.4 percentage points under a record-paired bootstrap. Across record-, dataset-, and domain-grouped splits, responses at $|α|=0.1$ are the strongest signal for outcomes at disjoint strengths $|α|\in\{0.25,0.5\}$. On held-out records, a predictor-driven selector chooses a coefficient or abstains, improves utility and reduces semantic-neighbor damage relative to a train-tuned fixed-strength policy, and avoids most evaluations in a dense scan. Across three residual-norm-matched base models, learned directions retain selective-path gains and low-dose responses yield 0.801-0.828 record-held-out macro AUROC. PML therefore turns memory localization into a falsifiable forecast of margin-level selective outcomes and a risk-aware intervention decision.

Related papers

• MemRiskBench: Trace-Aware Risk-Preserving Evaluation for Long-Horizon LLM Agents [5.605743575854166]
  MemRiskBench is a 120-episode scripted benchmark with full trace logging and no LLM-as-judge on the pass/fail path.<n>A risk-preserving subset selector preserves risk-type coverage and high-risk detection using trace-grounded deterministic features.<n>All episodes, traces, the scoring implementation, and the selector are released to support reproducible evaluation and risk assessment of deployed LLM agents.
  arXiv  Detail & Related papers  (2026-09-14T03:31:19Z)
• GeoID-PINN: Identifiability-Aware Regional Epidemic Inference with Geographic Coupling [1.0875426095207128]
  We introduce a physics-informed neural network (PINN) for susceptible-infectious-recovered-deceased dynamics.<n>We regularize a spatial prior constructed from distance, adjacency, commuting, or lead-lag information.<n>Relative to an autoregressive negative-binomial baseline, Forecast-Trained Geo-PINN reduces mean squared error (MSE) from 32,957 to 11,468 and mean absolute error (MAE) from 70.60 to 57.73.
  arXiv  Detail & Related papers  (2026-07-28T05:59:23Z)
• How Useful is Causal Invariance for Domain Adaptation in Finite-Sample Settings? [58.740078141879984]
  Machine learning models often degrade when they are deployed on a target distribution that differs from the source distributions they were trained on.<n>Recent work in causality-based domain generalization has shown how shared causal structure can induce invariant predictors.<n>In this paper, we study when full or partial causal knowledge can provably improve supervised domain adaptation.
  arXiv  Detail & Related papers  (2026-06-10T21:07:49Z)
• Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning [55.264863369127774]
  Current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.<n>We propose Hidden-Align, an auxiliary loss function that aligns the last-layer hidden states of correct rollouts at the anchor token during RL training, with zero overhead in both training and inference.<n>On eight mathematical reasoning benchmarks, Hidden-Align improves average pass@1 over the DAPO baseline by 3.8, 6.2, and 5.4 percentage points on Qwen3-1.7B, 4B, and 14B respectively, with consistent pass@k gains across all three scales.
  arXiv  Detail & Related papers  (2026-06-02T06:51:15Z)
• Clipping Bottleneck: Stabilizing RLVR via Stochastic Recovery of Near-Boundary Signals [83.0127582612634]
  Near-boundary Rescue (NSR) is a minimal, plug-and-play modification that retains out-of-bound tokens to recover lost signals.<n>NSR substantially improves training stability and delivers consistent gains over strong baselines such as DAPO and GSPO.
  arXiv  Detail & Related papers  (2026-05-21T16:45:31Z)
• Decision-Path Patterns as Tree Reliability Signals: Path-based Adaptive Weighting for Random Forest Classification [0.9607593117607193]
  Random forests construct each tree with a different, randomised representation of the feature space.<n>Their uniform voting cannot correct errors in regions where trees with incorrect representations outnumber correct ones.<n>This paper proposes using the structural pattern of each tree's decision path as a per-sample reliability signal.
  arXiv  Detail & Related papers  (2026-05-20T05:15:09Z)
• DISA: Offline Importance Sampling for Distribution-Matching LLM-RL [56.9445657766829]
  We introduce DISA, which moves this calibration problem outside the RL loop.<n> DISA draws proposal trajectories offline, estimates the partition function via importance sampling, and freezes the resulting partition-function estimate.<n>On two open-weight backbones across six math and three code benchmarks, DISA matches or exceeds the online-coupled distribution-matching baseline Flow.
  arXiv  Detail & Related papers  (2026-05-17T07:14:44Z)
• Harmful Intent as a Geometrically Recoverable Feature of LLM Residual Streams [0.0]
  Harmful intent is geometrically recoverable from large language model residual streams.<n>We characterise this geometry through six direction-finding strategies.<n>A direction fitted on AdvBench transfers to held-out HarmBench and JailbreakBench with worst-case AUROC 0.96.
  arXiv  Detail & Related papers  (2026-04-20T23:02:37Z)
• Backdoor Cleaning without External Guidance in MLLM Fine-tuning [76.82121084745785]
  Believe Your Eyes (BYE) is a data filtering framework that leverages attention entropy patterns as self-supervised signals to identify and filter backdoor samples.<n>It achieves near-zero attack success rates while maintaining clean-task performance.
  arXiv  Detail & Related papers  (2025-05-22T17:11:58Z)
• Ranking Free RAG: Replacing Re-ranking with Selection in RAG for Sensitive Domains [13.58151841630302]
  We propose a novel method METEORA that replaces re-ranking in RAG with a rationale-driven selection approach.<n>We show METEORA improves generation accuracy by 33.34% while using approximately 50% fewer chunks than state-of-the-art re-ranking methods.<n>In adversarial settings, METEORA significantly improves the F1 score from 0.10 to 0.44.
  arXiv  Detail & Related papers  (2025-05-21T20:57:16Z)
• ROPO: Robust Preference Optimization for Large Language Models [59.10763211091664]
  We propose an iterative alignment approach that integrates noise-tolerance and filtering of noisy samples without the aid of external models. Experiments on three widely-used datasets with Mistral-7B and Llama-2-7B demonstrate that ROPO significantly outperforms existing preference alignment methods.
  arXiv  Detail & Related papers  (2024-04-05T13:58:51Z)
• Divide and Contrast: Source-free Domain Adaptation via Adaptive Contrastive Learning [122.62311703151215]
  Divide and Contrast (DaC) aims to connect the good ends of both worlds while bypassing their limitations. DaC divides the target data into source-like and target-specific samples, where either group of samples is treated with tailored goals. We further align the source-like domain with the target-specific samples using a memory bank-based Maximum Mean Discrepancy (MMD) loss to reduce the distribution mismatch.
  arXiv  Detail & Related papers  (2022-11-12T09:21:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.