Develops a method to audit indirect feature influence in complex models.
problem Auditing indirect feature influence in complex, black-box models.
method Disentangled influence audits using disentangled representations.
result Can detect proxy features and show which ones affect model outcomes most.
Proposes a new model to capture joint influence of correlated events on user search behavior.
problem Real-world events influence each other and pose joint influence on user search behavior, not independent.
method Joint Influence Model based on Multivariate Hawkes Process.
result The model captures the temporal dynamics of joint influence and outperforms baseline methods.
Establishes statistical and computational bounds for influence diagnostics.
problem Identifying influential datapoints or subsets in machine learning models.
method Finite-sample statistical bounds and computational complexity for influence functions and approximate maximum influence perturbations.
result Established statistical and computational guarantees for influence diagnostics.
Dynamic Influence Tracker measures changing sample importance during model training.
problem Static influence measurements during training overlook how sample importance varies over time.
method Dynamic Influence Tracker (DIT) captures time-varying sample influence across arbitrary time windows.
result DIT reveals distinct learning phases with shifting priorities and detects corrupted samples more efficiently.
RelatIF selects more intuitive training examples for explaining model predictions.
problem Influence functions identify outliers as explanatory examples, leading to poor explanations.
method RelatIF separates global and local influence, optimizing for local relative to global effects.
result Examples selected by RelatIF are more intuitive than those from influence functions.
Influence functions are inaccurate in deep learning models, especially for deeper networks.
problem Inaccuracies in influence functions in deep learning models.
method Empirical study of influence functions in neural network models trained on various datasets.
result Influence estimates are often erroneous for deeper networks and require regularization.
Unified approach for influence maximization using diffusion cascade representations.
problem Influence maximization on networks with diffusion cascades.
method Multi-task neural network learning influencer and susceptible vectors; greedy algorithm for influence maximization.
result IMINFECTOR outperforms other methods in efficiency and seed set quality.
Proposes second-order influence functions for identifying influential groups in test-time predictions.
problem Identifying influential groups in test-time predictions for black-box models.
method Second-order approximations of the effect of removing a group of training samples on model predictions.
result Improves the correlation between computed influence values and ground truth values for linear models.
We tackle robust influence maximization in social networks with hyperparametric edge probabilities.
problem Maximizing worst-case influence in social networks with hyperparametric edge probabilities.
method Proposed a model with NP-hard proper robust optimization, using sampling and multiplicative weight updates.
result Empirically validated method outperforms state-of-the-art robust influence maximization techniques.
Paper presents method to audit black-box models for indirect feature influence.
problem Understanding how features indirectly influence model predictions in black-box models.
method Develops a technique to study feature influence without knowing model internals.
result Can find attribute influences even when features are not directly used by the model.
Influence functions explain black-box model predictions by tracing back to training data.
problem Explain predictions of black-box models.
method Use influence functions from robust statistics to trace predictions back to training data.
result Influence functions provide valuable insights into model behavior and training data.
DataInf efficiently approximates data influence in large models, improving transparency and identifying mislabeled data.
problem Efficiently estimating data influence in large-scale models like LoRA-tuned LLMs and diffusion models.
method DataInf uses a closed-form expression to approximate influence scores efficiently.
result DataInf outperforms existing methods in computational and memory efficiency, accurately identifying influential data points.
Influence functions help study large language model generalization, revealing surprising decay patterns.
problem Understanding and mitigating risks in large language models (LLMs).
method Eigenvalue-corrected Kronecker-Factored Approximation (EK-FAC) to scale influence functions to LLMs.
result Influences decay to near-zero when key phrases order is flipped, revealing a surprising limitation.
Maximizes influence in adversarial networks by choosing optimal seed nodes.
problem Maximizing influence in networks with an adversarial edge specification.
method Formulated as a repeated game between player and adversary, with online node selection.
result Established upper and lower bounds on minimax pseudo-regret.
The paper improves bounds on influence in network models using nonbacktracking walks.
problem Computing the expected influence in network models.
method Nonbacktracking walks, FKG inequalities, message passing.
result Improved bounds on influence computation in independent cascade models.
The paper extends influence functions to sequence tagging tasks for better model interpretability.
problem Lack of interpretability methods for sequence tagging models.
method Define and compute influence of training instance segments on test segment predictions.
result The segment influence method tracks with true influence and identifies annotation errors.
Improved scalability and interpretability in training data attribution.
problem Identifying which training data drives specific behaviors, especially unintended ones.
method Leveraging interpretable structures within the model to attribute model behavior to semantic directions, not individual test examples.
result Simple probe-based attribution methods are first-order approximations of Concept Influence that achieve comparable performance while being over an order-of-magnitude faster.
Enhances influence functions for deep models without costly Hessian inversion.
problem Challenges in computing Hessian matrix inverse for deep models.
method Substitutes Hessian inverse with identity matrix and extends applications.
result TracIn method performs well in measuring model utility and fairness.
Develops a faster model selection method using influence functions.
problem Time-consuming and computationally inefficient model selection.
method Model-specific data subsampling guided by influence functions.
result Empirically shows improved selection of high-quality models.
Paper introduces a multi-stage influence function to track model predictions.
problem Improving natural language processing and computer vision performance.
method Develops a multi-stage influence function to track predictions from finetuned models back to pretraining data.
result Identifies pretraining examples contributing most to finetuning task predictions.
We present the Bayesian Echo Chamber, a new Bayesian generative model for social interaction data. By modeling the evolution of people's language usage over time, this model discovers latent influence relationships between them. Unlike previous work on inferring influence, which has primarily focused on simple temporal…
Bayesian and ML approach for IM-RO problem in OSNs.
problem Estimating influence model parameters for maximizing revenue in OSNs.
method Bayesian hierarchical model with Naive Bayes, Decision Trees, and Random Forest classifiers.
result Efficient and practical method for estimating influence model parameters.
The paper simplifies influence computations for large-scale machine learning models.
problem Improving training efficiency and accuracy in large-scale models.
method Study influence functions, define memorization, simplify computations.
result Influence functions can be practical for large-scale models, indicating memorization.
New method estimates data influence efficiently by leveraging test samples.
problem Efficiently estimating influence of training data on model predictions.
method Mirrored Influence Hypothesis, forward pass for test samples.
result Significant improvement in efficiency over existing methods.
New framework for online influencer selection considering cost constraints.
problem Real-world social advertising budget limitations and cost variability of influencers.
method Introduces a budgeted framework for online influence maximization using an algorithm with semi-bandit feedback.
result Improves the state of the art regret bound for cardinality constraint setting.
I-GCN improves GCNs' robustness against adversarial attacks.
problem Adversarial attacks degrade GCNs' performance in security-critical applications.
method Influence mechanism divides node effects into introverted and extroverted influences.
result I-GCN achieves higher accuracy rates than state-of-the-art methods in defending against adversarial attacks.
Improves understanding of PWS by calculating influence of sources and data.
problem Understanding the influence of each component in PWS.
method Proposes source-aware Influence Function (IF) to decompose and calculate influence.
result Improves end model's generalization performance and identifies mislabeling.
Paper presents a new way to estimate model changes without full model evaluation.
problem Efficiently estimating changes in model parameters and outputs due to data point removal.
method Dual representation of influence functions for linearizable models, reducing computational complexity.
result The dual representation can be an efficient alternative to original influence functions, especially for large models.
New research investigates why influence functions are fragile and proposes new validation procedures.
problem Understanding and mitigating the fragility of influence functions in deep learning model explanations.
method Verification of influence functions using various conditions and procedures, including convexity and non-convexity.
result Validation procedures may cause the observed fragility of influence functions.
New algorithm reveals feature influence of machine learning models.
problem Causal influence measures can be misleading when applied to atypical data.
method Proposes an active learning algorithm to constrain influence measures.
result Trained models retain accuracy on in-distribution points while generalizing better to out-of-distribution points.
FIA method provides explainable recommendations for matrix factorization models.
problem Lack of explainability in latent factor models for recommendation.
method Influence functions from robust statistics to deliver neighbor-style explanations.
result FIA method successfully enforces explicit neighbor-style explanations to LFMs.
New algorithm estimates influence in social networks without sampling.
problem Estimating the number of influenced nodes in social networks.
method Dynamic message-passing approach for scalable influence estimation.
result Upper bound on expected spread for general graphs, exact for treelike networks.
A new framework maximizes influence spread in social networks by accounting for inter-community diffusion.
problem Real-world social networks have inter-community influence that is often overlooked in community-based IM approaches.
method Community-IM++ uses a heuristic based on community-based diffusion degree and progressive budgeting to model and prioritize cross-community diffusion.
result Community-IM++ achieves near-greedy influence spread at up to 100 times lower runtime than existing methods.
Study examines influence diagnostics in high-dimensional M-estimation.
problem Understanding influence diagnostics in high-dimensional settings.
method Characterized the distribution of leave-one-out influences in high-dimensional Gaussian M-estimation.
result The distribution of influences converges to a limiting measure in high-dimensional settings.
Captures data influence changes during training.
problem Traditional influence functions fail for modern training methods.
method Formalized trajectory-specific LOO influence, using data value embedding.
result Data influence varies by training stage, early and late stages have greater impact.
Shapley Homology measures sample influence on neural networks' manifold topology.
problem Assumption of iid samples simplifies manifold analysis in machine learning.
method Shapley Homology framework quantifies sample influence on neural networks' manifold topology.
result Higher influence scores correlate with greater impact on neural network accuracy.
Influence functions accurately measure group effects in large datasets.
problem Measuring the effect of large groups of training points in models.
method First-order Taylor approximation of influence functions, tested on various real-world datasets.
result Influence functions correlate well with actual group effects, even with large errors.
Model financial markets with social media influences using hierarchical networks.
problem Understanding social media's impact on financial markets.
method Agent-based model with hierarchical influence network.
result Model accurately simulates real-world financial market behaviors.
The article proposes a dynamic model for a company's life cycle under competitive influence.
problem Modeling a company's life cycle in a competitive environment.
method Utilized Markov model with known action costs and transition probabilities, affected by outside factors.
result Demonstrates the usefulness of the model in determining future actions of a company.
Algorithm maximizes influence in networks with limited feedback.
problem Maximizing influence in networks with limited feedback.
method Bandit algorithm using local node degree observations.
result Local observations are sufficient for maximizing global influence.
Study shows influence functions are poor for neural networks but useful for identifying influential examples.
problem Influence functions misalign with leave-one-out retraining in neural networks.
method Decomposed the discrepancy into five terms and studied their contributions across different architectures and datasets.
result Influence functions are a good approximation to the proximal Bregman response function (PBRF), useful for identifying influential examples.
Model predicts market dynamics of competing technologies.
problem Understanding market dynamics and technology competition.
method Modified Replicator Dynamics model with external influences.
result Shows how external influences affect market outcomes.
Efficiently selects seed nodes to maximize content influence in unknown social networks.
problem Maximizing content spread in social networks with unknown network model.
method Formulated as an infinite-horizon discounted MDP, uses model-based reinforcement learning to select seed users adaptively.
result Established a regret bound of O ~ ( T ) \widetilde O(\sqrt{T}) O ( T ) for the algorithm. Optimal sub-sampling method improves statistical model performance.
problem No well-motivated approach for non-uniform sub-sampling in statistical models.
method Asymptotically linear estimator and influence function.
result Optimal influence function based method outperforms previous approaches for linear regression models.
Better Hessian approximations improve influence function attributions in deep learning.
problem Influence functions are difficult to compute due to ill-conditioned Hessians, leading to poor data attribution performance.
method Investigated the impact of Hessian approximation quality on influence-function attributions in a controlled setting.
result Better Hessian approximations consistently yield better influence score quality.
Gradient rollback explains neural models by identifying influential training examples.
problem Explain predictions of neural black-box models, especially in applications requiring user trust.
method Gradient rollback, a general approach for influence estimation applicable to neural models with limited parameter updates.
result Gradient rollback provides faithful explanations for neural models, including knowledge graph embedding methods.
Influence sketching identifies influential samples in large datasets.
problem Finding important samples in large-scale regression models.
method Developed a scalable version of Cook's distance using random projections.
result Deleting high-influence samples reduces predictive accuracy significantly.
New IF method improves accuracy in deep neural networks with noisy data.
problem Inaccurate influence estimates in deep neural networks, especially with noisy data.
method Established a connection between influence estimation error, validation set risk, and sharpness, introducing a novel estimation form for flat validation minima.
result Our novel Influence Function approach provides more accurate influence estimates, validated across various tasks.