Overview

Gradient interference occurs when tasks or objectives produce updates that compete in shared parameters, potentially causing negative transfer, forgetting, or unstable training. Researchers study it in settings such as multi-task and continual learning, multimodal models, federated training, reinforcement learning, and physics-informed neural networks. Common approaches measure gradient alignment and use projection or reweighting methods, while others isolate updates with adapters, parameter subspaces, routing, or separate modules. A central tension is that directional conflict does not always predict actual performance harm, so interventions must account for context, gradient magnitude and variance, and trade-offs among objectives.

Latest papers

64 papers4 papers in the last four weeks, down 69% on the four weeks before. 0.0% of all new papers.

Oct 27, 2026

GradCO: Behavior-Aware Gradient-Conflict Offloading for Text-Based Person Anomaly Search

Text-based Person Anomaly Search (TPAS) retrieves a target pedestrian using natural-language descriptions of appearance and normal or anomalous behavior. Unlike conventional text-based person retrieval, TPAS must preserve identity cues while distinguishing fine-grained behaviors. It presents two challenges: same-identity, different-behavior hard negatives cause gradient conflict during granular alignment, while strong appearance cues can obscure weak behavioral evidence during aggregation. We propose GradCO (behavior-aware Gradient-Conflict Offloading), comprising three complementary components. Behavior-Aware Mixture-of-Experts (B-MoE) forms appearance, normal-behavior, and anomalous-behavior streams whose roles are learned from retrieval objectives without expert-specific semantic labels. Conflict Offload Learning (COL) uses pair-conditioned offload anchors in similarity computation to reduce the conflicting-gradient burden on the main text representations. Hierarchical Intent Alignment (HIA) aggregates expert similarities and features using query-conditioned weights. Experiments on five TPAS/TIPR benchmarks show GradCO significantly surpasses SOTA methods and validate our design.
Oct 22, 2026

Cautious Conflict-Free Inverse Gradient Methods

Partial differential equations (PDEs) are indispensable tools for modeling complex phenomena in physics and engineering. Physics-Informed Neural Networks (PINNs) have emerged as a powerful alternative to traditional numerical solvers by directly incorporating physical constraints into neural network loss functions. Despite their advantages, PINN training requires multi-objective optimization, which often leads to significant gradient conflicts among boundary, initial, and residual loss terms. The Conflict-Free Inverse Gradient (ConFIG) method addresses these conflicts; however, integrating ConFIG with adaptive optimizers such as Adam introduces a critical vulnerability: the conflict-free direction generated by ConFIG is further processed by the optimizer, and the resulting weight update may no longer be conflict-free with respect to the task gradients. To address this issue, Cautious ConFIG is proposed as a concise three-line extension to the ConFIG algorithm that empirically restores the conflict-freeness guarantee prior to the weight update. The method is evaluated on three PINN benchmarks: the 1D Viscous Burgers' equation, the 1D Nonlinear Schrödinger equation, and the 2D Kovasznay flow. Experimental results demonstrate consistent performance improvements over the standard ConFIG baseline. Additionally, the stabilizing effect of cautious masking and its reaction to conflict severity, task number, and optimizer lag are investigated using a simple toy system.
Oct 21, 2026

CORD: Conflict-Aware Optimization for Joint RL and Distillation in Agents

Joint reinforcement learning and self-distillation combines task rewards with dense teacher guidance for agent training, but their gradients can partly cancel. A mixing coefficient controls relative scale without selectively removing opposing components. We present CORD, a general framework for conflict-aware optimization of two objectives. CORD separates the choice of protected objective from the condition that triggers projection, and combines the gradients once per optimizer mini-batch. A nearest-point formulation yields a closed-form update and characterizes its first-order effect on the protected objective before optimizer transformations. Gradient diagnostics connect conflict in agent training to lost local optimization progress and motivate selective correction. As a gradient-combination mechanism, CORD can be added to different joint training recipes while retaining their objectives. Across three such recipes, it improves ALFWorld success by 1.5-15.9 percentage points. On WebShop, it improves two of the three pairs, with SDAR rising from 60.7% to 64.6% success. Our analyses explain why protection direction and projection gate must be chosen together, and why the distillation objective and its mixing weight remain consequential after projection.
Oct 16, 2026

Deconstructing Multi-Task Active Learning: The Paradox of Gradient Conflict and Orthogonal Decomposition

Existing multi-task active learning (MTL-AL) methods predominantly rely on linear scalarization of heterogeneous objectives such as uncertainty, diversity, and gradient conflict. We argue that this entangled formulation is fundamentally limited, as linear combination fails to capture Pareto-optimal trade-offs under conflicting objectives. We propose a shift from combination to deconstruction and introduce a principled framework that rethinks MTL-AL along three orthogonal dimensions. First, we identify the Conflict Paradox: although the optimization literature treats gradient conflict as interference to be suppressed, penalizing conflict when ranking candidates is measurably worse than not scoring it at all. We therefore decouple the two concerns: the acquisition function does not score conflict, and the resulting interference is resolved by gradient surgery during training. Second, we introduce Orthogonal Decomposition, separating acquisition into Informativeness and Exploration. This two-stage process, informativeness filtering followed by coverage maximization, prevents "diversity dilution" by excluding uninformative outliers before enforcing diversity. Third, we propose Cross-Task Mutual Information (CTMI), a covariance-based statistic of cross-task dependency computed from MC-dropout task entropies. Extensive experiments on COCO, Cityscapes, and NYUv2, including comparisons against recent MTL-AL and data-selection baselines, demonstrate consistent improvements over state-of-the-art methods, highlighting that structural rethinking of acquisition yields larger gains than incremental heuristic design. Code is available at https://github.com/Juho-Jung/multitask_active_learning_official.
Oct 10, 2026

MCPDisCo: Seesaw-Mitigated Tuning for MCP-Native Agents

The Model Context Protocol provides a shared interaction substrate for the agentic web, allowing heterogeneous services to expose capabilities through a standardized interface. Learning to use this ecosystem entails a hierarchy of capabilities: tool-format skills, server-level transfer, and cross-server composition. We present MCPDisCo, an ecosystem-scale reinforcement learning framework for MCP-native agents. Using a Thought-Markov Decision Process (Thought-MDP), the framework explicitly represents internal reasoning alongside routing, execution, and termination and trains on real-server, long-horizon interactions spanning 70 MCP servers and 527 tools across six domains. We study the reasoning--tool seesaw, a joint-learning failure mode in which improvement in one objective can accompany deterioration in the other. To mitigate potential interference, MCPDisCo combines token-masked reward attribution with bidirectional gradient surgery: the former separates direct supervision for reasoning and action spans, while the latter removes conflicting components from both objective gradients in shared parameters. MCPDisCo achieves higher success with fewer chain-of-thought tokens. On the out-of-distribution MCP-Atlas and the GAIA benchmark, MCPDisCo maintains monotone gains, confirming that the learned behavior generalizes beyond the training server distribution and broader agentic tasks. Model weight and implementation are available.
Oct 10, 2026

AcrossVAM2.0: Conflict-Aware Point Supervision for Parameter-Efficient Video Diffusion Adaptation

Explicit point correspondence is attractive for adapting video generators: it describes how visible surfaces move, but can overwrite the appearance prior learned by a large pretrained model. We study this tension in a 5.01B-parameter text-to-video diffusion transformer. ACROSSVAM2.0 combines a confidence-weighted track-and-visibility head, zero-initialized residual feedback, and asymmetric PCGrad that removes only the auxiliary component opposing the primary flow-matching gradient on shared LoRA parameters. The design preserves the generator exactly at initialization and protects its primary update during adaptation. In a four-way ablation trained for 1,000 steps with three independent seeds, the full method lowers KLT forward-backward cycle error in all three seeds relative to a matched flow-only continuation (mean seed-level difference -0.00711, SD 0.00505), while robust sharpness remains positive in all three. An expanded 24-prompt evaluation (216 videos) yields 9.7% lower aggregate CoTracker3 cycle error than the matched control and near-matched performance on six VBench dimensions; however, the seed-level tracking effect is heterogeneous and I3D-FVD regresses from 0.1824 to 0.2022. These results support a bounded claim: conflict-aware adaptation improves correspondence consistency without broad perceptual degradation, but does not uniformly improve video quality. We release equations, pseudocode, matched outputs, source tables, and artifact provenance.
Oct 9, 2026

Beyond Domain-Routed Supervision: Cross-Teacher Guarding for Multi-Teacher On-Policy Distillation

Multi-teacher on-policy distillation (MOPD) extends on-policy distillation (OPD) by routing domain-specific data to specialized teachers, integrating their capabilities through dense token-level supervision on student-generated trajectories. However, domain routing only distinguishes the supervision sources for different domains; updates from all domains still act on shared parameters. When teachers differ substantially in response style and domain expertise, supervision that is effective in isolation may still lead to persistent gradient conflicts and capability degradation during joint optimization. To address this challenge, we propose Cross-Teacher Guarding for MOPD (Road MOPD), which introduces non-routed teachers as auxiliary cross-domain critics on the same student trajectories and coordinates multi-teacher supervision through token-level credit assignment and gradient updates. Specifically, cross-teacher-guided credit assignment uses teacher agreement to strengthen token-level corrections jointly supported by the teachers and attenuate conflicting signals; preventive multi-teacher gradient projection uses auxiliary gradients from the same trajectories to constrain potentially conflicting updates and adaptively controls projection strength to retain target-domain learning signals. In experiments, Road MOPD mitigates degradation in the disadvantaged domain, gains over 5 percentage points under high conflict, and improves overall performance in low-conflict and three-domain settings.
Oct 7, 2026

SGF+: Decoupling Gradient Flows for Autoregressive Video Generation

Autoregressive video generation requires denoising the current frames while writing their key-value representations as context for future predictions. However, these two roles typically share parameters, and we find that their gradients exhibit distinct patterns and systematic negative alignment, hindering the joint optimization of visual quality and temporal consistency. We introduce Self Gradient Forcing Plus (SGF+), which assigns separate parameters to context writing and denoising while preserving their interaction through causal attention. Both roles are jointly optimized using the original generation objective without auxiliary losses, with context writing supervised through its contribution to future predictions. This simple change improves visual quality and long-horizon consistency over the evaluated baselines in both framewise and chunkwise generation, without additional video training data or a longer training horizon. Trained on only 5s rollouts, SGF+ supports continuous generation for up to 24 hours without long-video fine-tuning. These results highlight role-specific parameterization as an effective design principle for high-quality autoregressive video generation and native long-horizon extrapolation.
Oct 6, 2026

Which and When to Admit: Gradient Admission for Data-Centric Small Language Model Finetuning

LoRA fine-tuning adapts small language models (SLMs) to heterogeneous instruction data within a low-rank update subspace, making it vulnerable to three structural problems: conflicting gradients that cancel, static data selection that cannot track evolving learning dynamics, and subspace saturation that causes later updates to overwrite useful directions. We argue that effective adaptation therefore requires controlling which data-induced gradients enter the LoRA subspace and when. We propose GRADE (GRadient-Aligned Data-centric rEcipe), a data-centric framework combining two mechanisms: a state-aware selector that continually admits samples aligned with the evolving multi-task gradient field, and a self-calibrating step-level gate that rejects updates likely to cause destructive overwrite near saturation. Across three current-generation backbones and a heterogeneous seven-dataset instruction pool, GRADE outperforms strong data-selection and PEFT-stabilization baselines in accuracy and robustness. It is the only method to improve consistently over standard LoRA on every architecture, while producing more coherent gradient trajectories and less destructive overwrite. These results show that successful SLM adaptation depends not only on which data are selected, but also on which gradients are allowed to enter and persist in the constrained update subspace.
Sep 30, 2026

DAMPER: Return-Prioritized Gradient Control for Smooth Policies

Actor-critic methods achieve strong performance in continuous control, but their policies can produce highly oscillatory actions. A common remedy is to add auxiliary smoothness losses. However, their contribution can be negligible when their gradients are small relative to the native actor gradient. Moreover, existing methods often combine multiple auxiliary losses, complicating loss balancing without necessarily improving the return-smoothness trade-off. We introduce DAMPER (Direction-Aware Magnitude-Controlled Projection with Explicit Return Priority), which combines the native actor gradient with a temporal-consistency gradient through conflict-conditioned projection and adaptive magnitude control. It removes the auxiliary component opposing the actor gradient and scales the retained temporal direction relative to the actor gradient norm, preserving positive alignment with the native actor gradient. Experiments with TD3 and SAC on six continuous-control tasks show reduced action oscillation relative to the native agents in all 12 task-backbone pairs and the best oscillation score among the compared methods in eight, with task-dependent return trade-offs.
Sep 29, 2026

Beyond Layers: Position-Resolved Gradient Conflict and Position-Aware Modulation for Unified Multimodal Models

Unified multimodal models (UMMs) train image understanding and autoregressive image generation on shared parameters, and the two objectives are known to interfere. Existing diagnoses and remedies operate at the resolution of layers or experts, measuring conflict per layer and resolving it by separating parameters. We argue that this resolution hides an orthogonal axis. Generation in a UMM is next-token prediction over a raster sequence of visual tokens whose roles vary systematically with position, so how strongly a generation gradient interferes with understanding should depend on where in the sequence it originates. We introduce a position-resolved interference map that attributes understanding-generation gradient conflict to visual-token positions within every layer, computed from a single backward pass at 1.2×1.2\times the cost of a standard backward pass. On Show-o and Janus-Pro, position explains a large share of conflict variance after controlling for depth (partial η2=0.31η^2=0.31 vs. 0.350.35 for layer on Show-o; 0.150.15 vs. 0.300.30 on Janus-Pro): the first quarter of the sequence has a mean gradient cosine of −0.18-0.18 against understanding, the last quarter −0.02-0.02. The dependence survives per-position gradient-norm normalization, retaining 8080% of its effect size, and conflict strength tracks semantic content (Spearman ρ=0.64ρ=0.64). Building on the map, we propose position-aware modulation (PAM), which removes the anti-aligned component of generation gradients only at high-conflict positions without changing the architecture. Under a matched trainable-parameter budget, PAM improves over layer-wise separation by +21+21 MME and +2.4+2.4 GenEval points on Show-o while matching it on POPE and overall FID; a random-position control recovers about 3131% of the gain. Position-based and layer-based separation are complementary degrees of freedom and can be combined.
Sep 29, 2026

Does Gradient Conflict Predict the Understanding--Generation Trade-off? A Controlled Audit of Conflict-Metric Validity in Unified Multimodal Models

Unified multimodal models (UMMs) are increasingly designed around gradient conflict between understanding and generation objectives. The premise that reducing these metrics improves the downstream understanding-generation trade-off has never been tested directly. We audit it in a controlled testbed, GRIDUMM, which mirrors key structural ingredients of UMM training while making the ground-truth trade-off exactly computable. Across 63 configurations and 372 measured checkpoints, no directional conflict metric reaches an absolute Spearman correlation of 0.3 with a confidence interval excluding zero for conflict measured during training against the eventual trade-off. A dose-response intervention that monotonically suppresses conflict leaves the trade-off flat, separating correlation from causation. The norm ratio is a generation-failure detector and becomes null among configurations that master generation. Functional interference measures outperform directional conflict metrics, while training loss tracks the trade-off strongly. Our results do not show that conflict is useless; they show that its validity as a diagnostic target must be established, not assumed, and we release the audit protocol as a reusable standard.
Sep 28, 2026

ORPG: Reconciling Multiple Reward Objectives through Objective-wise Policy Gradients

Multi-reward policy optimization requires a joint update that reflects both the learning signals and the intended relationships among objectives. We introduce Objective-wise Reconciled Policy Gradient (ORPG), which constructs a separate clipped policy objective for each reward and reconciles the resulting gradients into one policy update. For compatible gradients, a cosine-dependent interpolation coordinates their contributions through a partially normalized reference while preserving the norm of their sum. We characterize this update as the unique solution of a spherical directional compromise. For conflicting gradients, projection follows the task's priorities. We evaluate the same compatible rule in helpfulness--safety alignment and correctness--cost optimization for mathematical reasoning. ORPG substantially improves average Useful and Harmless scores over the strongest external baseline on each axis. In mathematics, it achieves the highest average full-budget accuracy and three-budget hypervolume among the compared methods, with more accurate and shorter responses than the initial policy. Component comparisons and training dynamics show the larger contribution of compatible coordination and a complementary benefit from conflict handling. These results support gradient reconciliation for objectives with equal standing and for objectives with an explicit priority.
Sep 28, 2026

When Sparse Reward Meets Dense Distillation: Training Dynamics of On-Policy Distillation

Reinforcement learning with verifiable rewards provides a sparse post-training signal: a single binary outcome evaluates the entire rollout, and every token receives the same sequence-level advantage regardless of its individual contribution. To complement this sparse supervision, a growing family of methods adds a scalar-weighted teacher KL term to the policy-gradient objective, providing dense token-level guidance that may be unreliable at some positions. Despite the benefits of combining these signals, their interaction during optimization can destabilize joint training. To understand how this instability develops, we study the learning dynamics of hybrid reward--distillation training through a neural tangent kernel (NTK) analysis. We introduce the cross-signal NTK KDR(n)K_{DR}(n), a token-level statistic that measures the alignment between reward and distillation gradients at position n. Through this analysis, we identify two failure modes: 1 Magnitude drowning, where the reward gradient exceeds the distillation gradient by orders of magnitude, so that even weak directional conflict can cause the distillation loss to rise despite its explicit inclusion in the training objective; and 2 Localized directional conflict, where the sequence-level advantage and the teacher's position-specific distribution induce opposing updates at the same token (KDR(n) ⁣< ⁣0K_{DR}(n)\!<\!0). The severity of these effects depends on the optimization regime: the gradient-norm ratio κ ⁣= ⁣∥∇LR∥/∥∇LD∥κ\!=\!\|\nabla\mathcal{L}_R\|/\|\nabla\mathcal{L}_D\| varies by roughly an order of magnitude across tasks, and our experiments reveal an empirical threshold beyond which naive mixing can lead to persistent training collapse. Motivated by these findings, we introduce the M3 family, which combines magnitude normalization with three strategies...
Sep 28, 2026

PMOPD: Task Ordering, Cycling, and Parameter-Update Subspace Protection in Multi-Teacher On-Policy Distillation

Multi-teacher on-policy distillation (MOPD) has emerged as a popular post-training paradigm for integrating specialized capabilities in frontier language models. Existing OPD research has primarily focused on optimizing single-task distillation through objective design, distillation scope, and teacher signal construction, whereas MOPD must aggregate multiple capabilities in shared parameters and address the resulting capability seesaw, in which improving one domain suppresses capabilities acquired from another. Inspired by the distinctive update geometry of OPD, we find that parameter updates from different tasks rapidly concentrate in their respective low-dimensional subspaces during MOPD, providing a direct geometric basis for identifying and controlling cross-task interference. We therefore propose PMOPD (Projection-based Multi-Teacher On-Policy Distillation), which constructs subspace memories from the cumulative parameter displacements of different tasks and projects both gradients and optimizer updates to remove components that interfere with protected task directions. We further develop a lightweight conflict probe to characterize task interactions and guide task ordering, together with a cycling strategy that balances subspace estimation and timely task revisitation. Experiments on representative Code, Reason, and Math tasks show that PMOPD improves every evaluated capability over MOPD, raising the average score across the three tasks by 2.54 points on Qwen2.5-7B and 2.09 points on Llama-3.1-8B. These consistent gains establish geometry-aware optimization as an effective and transferable approach to balanced multi-teacher distillation.
Sep 27, 2026

DGS-MLDG: Domain Gradient Surgery Guided Meta-Learning for Domain Generalization in Speech Deepfake Detection

Speech deepfake detection faces significant challenges due to domain shifts. Domain generalization (DG), particularly meta-learning for domain generalization (MLDG), offers a promising solution by simulating and mitigating domain shifts. However, MLDG is often hindered by conflicting gradients between its meta-train and meta-test objectives, leading to suboptimal performance. To address this problem, we propose domain gradient surgery (DGS), a meta-learning method that resolves conflicts through an asymmetric projection strategy. DGS removes the destructive component from the meta-test gradient, ensuring a conflict-free optimization trajectory versus the meta-train gradient. Furthermore, we introduce layer-wise DGS (LW-DGS), an efficient variant of DGS that dynamically identifies and intervenes only conflict-prone layers. Extensive experiments on challenging benchmarks demonstrate that DGS-MLDG and LW-DGS-MLDG achieve an average relative EER reduction of 5.29% and 4.04%, respectively.
Sep 25, 2026

Gradient Surgery for Physics-Informed Neural Networks

Physics-Informed Neural Networks (PINNs) are trained by optimising a composite objective that combines data fitting with physics-based constraints, typically resulting in a highly imbalanced multi-task optimisation problem. Under these conditions, existing optimisation strategies are affected by conflicting task gradients, leading to slow convergence and unstable training, particularly for stiff and high-frequency partial differential equations. We analyse gradient conflicts throughout training of PINNs with standard optimiser and investigate Multi-Task Deep Learning (MTDL) optimisation methods. In our analysis across four benchmark problems we observed that PINN optimisation exhibits three distinct phases in which angle- and magnitude-based gradient conflicts alternate, with only one present at a time. Building on these observations, we propose PAM-GS, a physics-aware gradient surgery method that adaptively mitigates task interference during training according to the observed conflict types. Experiments on four representative PDE benchmarks demonstrate that PAM-GS combines competitive solution accuracy with consistently strong task-balanced performance, outperforming existing methods on most problems.
Sep 13, 2026

Tackling Failure Modes of PINNs and PIKANs Using Conflict-Free Gradients

Scientific machine learning methods such as physics-informed neural networks (PINNs) increasingly rely on domain decomposition for better scalability while solving partial differential equations (PDEs) over complex geometries, yet the resulting composite loss comprising residual, boundary, and interface terms is highly susceptible to conflicting gradients that degrade training. This work bridges domain decomposition with projection-based gradient surgery to systematically mitigate such conflicts in 2D and 3D settings. We evaluate two existing projection-based algorithms, PCGrad and ConFIG, and identify their performance degradation in specific scenarios such as 3D domains with multiple overlapping interfaces. To address this limitation, we propose Norm-PCGrad, a normalized variant that achieves state-of-the-art accuracy across a range of 2D and 3D domain decomposition problems. Across the benchmarks considered, Norm-PCGrad consistently achieves the lowest relative L2L_2 error compared to training without gradient surgery as well as to existing algorithms such as PCGrad and ConFIG, while incurring negligible additional computational overhead. To improve computational efficiency of domain decomposition frameworks such as Extended PINN (XPINN), we propose replacing vanilla PINNs in selected subdomains with separable architectures such as Separable PINN (SPINN), reducing the computational cost from quadratic (or cubic) to linear. We additionally demonstrate that gradient surgery extends to physics-informed Kolmogorov-Arnold Networks (PIKANs), yielding substantial accuracy improvements for 3D domain decomposition and confirming the generality of the proposed approach across network architectures.
Sep 8, 2026

Not All Variables Agree: Reliability-Aware Variable-Wise Gradient Surgery for Multivariate Time-Series Forecasting

In data-driven training, multivariate time-series forecasting is usually optimized with a scalar loss averaged over samples, variables, and horizons. This averaging is convenient, but the optimizer sees only the aggregated gradient, which does not reveal whether the variable-wise contributions align or oppose one another. To quantify how often this disagreement arises, we measure the variable-wise gradients directly and find that 30.6% of their pairwise cosine similarities are negative on average across seven datasets. However, conflict and harm are not the same thing. Under shared training 35 of the 64 variables do worse than a full-input single-target oracle, and the harmed fraction is not reliably predicted by how often gradients conflict. We propose Per-Variable Surgery (PV-Surgery), an optimizer-side training strategy for backbones with cache-compatible layers. One backward pass builds variable-wise gradient proxies from output-side signals and keeps the pointwise forecasting loss. Reliability-aware selection targets layers whose proxy sums closely approximate their shared-gradient slices. Conditional pooling forms anchor and conflict pools without dropping variables. Common-direction surgery aligns variable or pooled gradients with their normalized mean and restores input norms to avoid reweighting. In experiments across five backbones, seven datasets, and four horizons, PV-Surgery lowers MSE by 3.61% and MAE by 2.93% on average. For multivariate forecasting, this indicates that the variable-wise structure hidden by mean-loss training is a usable optimization signal.
Sep 8, 2026

A Better Spur Should Start From Each Objective

Real-world Multi-Objective Reinforcement Learning (MORL) often suffers from sparse rewards, reward conflicts, and late-stage reward tug-of-war, causing traditional linear scalarization to experience severe metric oscillations. To address optimization conflicts among multiple objectives in real-world deployment scenarios, we propose Multi-Marginal Preference Optimization (MMPO), a fine-grained framework that intervenes at the data, gradient, and constraint levels rather than relying on coarse-grained global scalarization. Specifically, MMPO performs exposure debiasing to mitigate sparse and biased rewards, applies priority-aware orthogonal projection to decouple conflicting gradients, and introduces self-prompted gradient constraints to prevent dominant objectives from overwhelming weaker ones. Experiments on real-world e-commerce datasets show that MMPO improves training stability and consistently achieves better performance across conflicting metrics. Moreover, it generalizes robustly to broader tasks such as ToolRL and code generation, demonstrating its effectiveness as a practical paradigm for multi-objective alignment.
Sep 6, 2026

Selective Knowledge Control for Continual GUI Agent Learning over Application Streams

Continual learning is a crucial capability for Graphical User Interface (GUI) agents to adapt to evolving applications while retaining knowledge acquired from previous applications. Such application streams pose a challenging knowledge modeling problem: new applications often share underlying knowledge with past ones, yet also introduce specific knowledge that must not interfere with historical knowledge. In this paper, we propose activation-conditioned selective knowledge control, a lightweight method that achieves selective knowledge retention via neuron-level gradient manipulation. Our method maintains a compact historical knowledge state to protect highly activated MLP neurons that preserve previous knowledge. When a new application arrives, it performs real-time gradient surgery conditioned on forward activation. Concretely, the protected neurons are categorized into two types: unactivated neurons holding specific knowledge, whose gradients are truncated to prevent interference; and activated neurons holding shared knowledge, whose gradients are orthogonally projected to preserve stability while enabling adaptation. After each application stage, newly identified critical neurons are merged into the historical state for future learning. Empirical evaluations on multi-app sequential benchmark demonstrate that our method effectively mitigates catastrophic forgetting on prior applications while sustaining robust adaptation to new ones.
Sep 1, 2026

Gradient-Update Mismatch: Rethinking Conflict-Free Training of Physics-Informed Neural Networks

Training Physics-Informed Neural Networks (PINNs) requires jointly optimizing physics residual and initial/boundary condition loss terms, which often induce conflicting gradients. Gradient surgery methods mitigate this issue by constructing directions from loss-specific gradients to reduce conflict before optimizer transformation. However, even when the constructed direction is conflict-free, this property may not be preserved after optimizer transformation. Let ata_t denote the direction constructed by gradient surgery, utu_t the optimizer proposal, and Ct\mathcal{C}_t the conflict-free cone induced by the loss-specific gradients. We show that modern optimizers can transform ata_t through mechanisms such as historical state, adaptive scaling, preconditioning, or decoupled weight decay, so at∈Cta_t \in \mathcal{C}_t does not generally imply ut∈Ctu_t \in \mathcal{C}_t. We refer to this optimizer-induced discrepancy in conflict-freeness between ata_t and utu_t as Gradient-Update Mismatch (GUM). Accordingly, we propose Gradient-Update Alignment (GUA), which projects utu_t onto Ct\mathcal{C}_t to obtain the aligned update ptp_t and applies ptp_t to the parameters. When the optimizer maintains internal state, GUA further adjusts this state toward targets reconstructed from the applied update. We conduct extensive experiments and find that GUM is widespread across momentum, adaptive, and curvature-based optimizers, with conflict rates reaching up to 86.3%. Across all PINN settings, GUA achieves conflict-free applied updates and consistently improves various gradient surgery methods, reducing the relative L2L_2 error by up to 98.2% in individual settings. Data and code are available at https://github.com/JingXiao10/GUA.
Aug 31, 2026

PRIME: Mitigating Subgroup Optimization Competition in Shared CTR Top Networks with Plug-in Residual Input-Conditioned Mixture of Expert

Click-through rate (CTR) models vary in feature-interaction design, yet their top networks usually remain a single multilayer perceptron shared by all examples. Heterogeneous user, item, and context subgroups therefore update the same parameters; weakly aligned learning signals make the aggregate gradient a compromise among competing directions. We study the competition on Avazu with 4 models and 4 semantic fields. Across all architectures, semantic subgroups show lower Top-NN gradient cosine similarity than random groups matched by sample size and label ratio, with reductions of 0.23-0.37. This competition motivates input-conditioned experts, but directly replacing an established Dense mapping changes its initial function, sharing pattern, and capacity, obscuring the source of gains. We introduce PRIME (Plug-in Residual Input-conditioned Mixture of Experts), a Dense-anchored mixture of low-rank residual experts. PRIME anchors the original prediction and uses zero-residual initialization to match the Dense baseline exactly at training onset. Input-dependent routing weights low-rank experts for example-specific logit corrections; multi-bag aggregation and EMA load biases stabilize conditional estimation. We evaluate PRIME on held-out Avazu and Criteo test sets across 13 CTR architectures and five paired seeds. Median paired AUC gains are +0.0022 and +0.0066, with LogLoss reductions of 0.0011 and 0.0081, respectively. On FiBiNET and DCNv2, PRIME outperforms APG in all ten seed-level AUC comparisons while using fewer parameters and lower inference latency on both backbones. These results show that function-preserving conditional residuals add input-dependent capacity while preserving the Dense path and its optimization stability. Code is available at https://github.com/YH-learning/PRIME.
Aug 25, 2026

Absorbing Gradient Conflicts: Modeling Semantic Variance via Kent Distributions for Cross-Modal Hashing

Supervised proxy-based deep cross-modal hashing has become the dominant paradigm for large-scale retrieval. However, prevalent methods model class proxies as deterministic points in the embedding space. This rigid assumption causes severe gradient conflicts in multi-label scenarios, where gradient conflicts arising from label co-occurrence lead to severe gradient contention and optimization collapse. To resolve this, we propose Kent-based Distributional Proxy Hashing (KDPH), a novel framework that shifts proxy representation from static points to flexible anisotropic Kent distributions on the hypersphere. Unlike point proxies that must shift their positions to accommodate conflicting gradients, KDPH absorbs these conflicts by dynamically adjusting its directional variance. This allows the proxy to maintain a stable semantic mean direction while stretching to cover diverse label correlations. Furthermore, to ensure stable training of these geometric parameters, we derive a tailored loss function incorporating the Cayley transform to enforce strict orthogonality. To the best of our knowledge, KDPH is the first framework to successfully introduce the Kent distributions into cross-modal hashing. Experiments on three benchmark datasets demonstrate that KDPH mitigates proxy collapse and chaotic oscillation, significantly outperforms state-of-the-art methods. Code is available at https://github.com/Senmo996/KDPH-official-code.
Aug 24, 2026

CAT-GS: Balanced Multimodal Learning via Calibrated Gating and Fusion Surgery

End-to-end training of multimodal neural networks often exhibits unstable neural dynamics characterized by three coupled failure modes that degrade learning: (i) modality imbalance, where one branch dominates gradient-based optimization; (ii) unstable gating, where noisy confidence cues induce erratic modality selection; and (iii) fusion interference, where modality-specific gradients conflict at the shared fusion layer. We propose CAT-GS (Calibrated, Adaptive, Thresholded Gating with Fusion Surgery), a neural dynamics-based optimization controller for intelligent computing applications. CAT-GS operates during backpropagation without modifying model architectures, fusion modules, or task losses. Through calibration of teacher-derived reliability via temperature scaling and EMA smoothing, CAT-GS stabilizes neural dynamics using a margin-thresholded policy to switch between warm-up dropout, weak-modality prioritization, and weak-biased blending, stabilizes gradient magnitudes under aggressive gating via capped gradient-budget renormalization, and applies fusion-only PCGrad to reduce destructive cross-modal interference at the primary shared bottleneck. We evaluate CAT-GS on audio--visual multimodal pattern recognition benchmarks (CREMA-D, AV-MNIST, and VGGSound), a tri-modal setting (UR-FUNNY), controlled synthetic data (CG-MNIST), and additional cross-domain benchmarks (AVE and CMU-MOSI). CAT-GS improves or matches fused multimodal accuracy against strong imbalance-aware baselines (including OGM-GE, G2^2D, and UMT) across settings, and yields smoother gating behavior with fewer conflicting fusion gradients.
Aug 18, 2026

GUPO: Gradient Uncertainty-aware Policy Optimization for Post-Training Large Language Models

Group Relative Policy Optimization (GRPO) has become a widely used approach for post-training Large Language Models (LLMs) for reasoning. In GRPO, the group gradients induced by different queries within the same mini-batch are directly averaged to form the policy update. However, these group gradients can point in conflicting directions. Our empirical analysis suggests that group-gradient conflicts tend to be associated with less effective policy updates, motivating the need for a reliable aggregated update direction under such conflicts. Standard GRPO aggregation treats the realized group gradients as deterministic contributions and does not account for differences in their reliability during aggregation. To address this issue, we propose Gradient Uncertainty-Aware Policy Optimization (GUPO), which models each group gradient as a random variable under a Bayesian formulation and estimates its probability distribution. GUPO then derives gradient uncertainty using a Dirichlet-based formulation and uses it to calibrate the contribution of each group gradient during aggregation. Extensive experiments on multiple benchmarks demonstrate the effectiveness of GUPO.
Aug 4, 2026

SFT Conflicts, RL Coexists: A Theoretical and Empirical Analysis of Multi-Task Learning for LLMs

Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) exhibit fundamentally different behaviors in enhancing multi-task reasoning for large language models (LLMs). Our preliminary experiments revealed a phenomenon: SFT suffers from severe task conflicts under multi-stage training, whereas RL enables stable coexistence across diverse tasks. Empirically, we trace this to the parameter level, observing that RL induces sparse and approximately orthogonal updates across tasks. We provide a theoretical explanation for this mechanism by analyzing multi-task gradient interference. Our results reveal a distinction: interference in SFT is norm-limited, scaling with the absolute gradient magnitude, whereas interference in RL is variance-limited, bounded by the gradient variance induced by advantage normalization and on-policy optimization. This small variance bound yields near-orthogonal optimization directions across tasks. Leveraging this insight, we propose Parallel-RL, a paradigm that decouples multi-task training, significantly improving efficiency and flexibility.
Aug 3, 2026

CLEAR: Conflict-aware Learning via Evidence-guided Adaptive Routing for Unified Sparse-View 3D Gaussian Super-Resolution

Sparse-view 3D Gaussian Splatting Super-resolution is highly challenging since the sparse and low-resolution (LR) inputs lack sufficient geometric and high-frequency information for accurate reconstruction. To achieve high-quality reconstruction, existing sparse-view super-resolution methods adhere to two-stage pipeline that performs LR Gaussian reconstruction and then high-resolution (HR) Gaussian refinement, which directly results in stage-wise Gaussian transfer and reconstruction error accumulation. To this end, we propose CLEAR, a Conflict-aware Learning via Evidence-guided Adaptive Routing, as the first unified single-stage framework for Sparse-view 3D Gaussian Splatting Super-resolution. Specifically, CLEAR performs joint the optimization of authentic LR observations and external HR priors within a unified Gaussian representation. To mitigate the gradient conflicts introduced by sparse supervision during training, we propose a Gaussian-wise conflict-aware optimization strategy that regards the LR gradient as a reliable anchor and applies evidence-conditioned soft correction only to severe HR conflicts. Moreover, to recover high-frequency details, we introduce an evidence-guided Patch-to-Gaussian routing mechanism which estimates patch reliability and detail demand, lifts them into Gaussian space, and selectively routes high-frequency gradients and densification. Finally, we employ shared Gaussian dropout and a detached mid-training anchoring to enhance the robustness of training framework. Extensive experiments on both synthetic and real-world 4×4\times super-resolution benchmarks demonstrate that CLEAR consistently achieves state-of-the-art rendering quality and superior geometric fidelity.
Jul 27, 2026

Lantern: Conflict-Aware Gradient Blending for Physics-Guided Diffusion Models in Calorimeter Simulation

Monte Carlo simulation of calorimeter showers is a principal bottleneck for the High-Luminosity LHC, and diffusion models have emerged as fast, high-fidelity surrogates. Their denoising objective is purely statistical, however: a model can minimize it while placing the physics wrong. Existing physics-informed generative methods cannot close this gap, because they assume a closed-form law, a governing PDE residual or a hard per-sample constraint, that a shower does not supply: no per-sample PDE governs a stochastic cascade, and energy conservation fixes only one scalar per shower. Standard metrics ignore the correlation structure across calorimeter layers and voxels, comparing showers only in a physics feature space. We address both gaps. We introduce the Correlation Frobenius Distance (CFD), a single normalized score for correlation fidelity at layer-wise and voxel-wise scales. We then encode the soft per-sample structure available in a shower as two physics-aware auxiliary losses: a variance-stabilized voxel residual loss grounded in counting statistics, and a graph Laplacian loss over the detector geometry. We combine both with denoising through GradBlend, which anchors the step magnitude to the denoising gradient while letting the auxiliary steer its direction, yielding Lantern, a physics-guided diffusion surrogate. On CaloChallenge Dataset 2, injecting the physics losses through task-symmetric rules such as PCGrad, GradNorm, IMTL-G, and ConFIG inflates FPD by 2-100x relative to denoising alone, whereas GradBlend admits the same signal without regression and, with the Laplacian loss, Lantern improves both FPD and CFD. Our ablation on the auxiliary loss scheduler shows that the voxel residual loss, whose gradient conflicts with denoising, requires a terminal denoising-only phase to preserve shower fidelity, whereas the non-conflicting Laplacian loss is insensitive to the schedule.
Jul 20, 2026

FlexiGrad: Adaptive Gradient Modulation for Hierarchical Fine-Grained Classification

Many fine-grained recognition tasks contain hierarchical labels such as order, family and species. Although this supervision should be beneficial, jointly optimising all levels often leads to unstable training because coarse and fine classifiers impose inconsistent gradients on the shared backbone. This hierarchical gradient conflict prevents the model from learning a coherent coarse-to-fine representation. In this paper, we propose FlexiGrad, a simple and parameter-free method that regulates gradient interactions during backpropagation. FlexiGrad removes only the harmful conflicting component when tasks disagree and reinforces the shared direction when they partially agree through a smooth hierarchy-aware weighting function. This produces stable optimisation and preserves both global structure and fine-grained discriminative cues. FlexiGrad integrates into existing architectures without modification while improves multi-granularity accuracy on CUB-200-2011, FGVC-Aircraft and Stanford Cars. The code will be available at PRIS-CV/FlexiGrad.