Study non-oblivious adversarial bandits with delayed feedback and propose algorithms with improved regret bounds.
problem Adversarial bandit problem with delayed, composite anonymous feedback.
method Propose wrapper algorithm for non-oblivious delay setting, achieving o ( T ) o(T) o ( T ) policy regret. result Achieve o ( T ) o(T) o ( T ) policy regret for many adversarial bandit problems with bounded memory loss sequences. New algorithm achieves both static and dynamic regret optimally against an oblivious adversary for deterministic losses.
problem Achieving optimal static and dynamic regret simultaneously in adversarial bandits.
method Extends impossibility result to deterministic losses, uses negative static regret and Blackwell approachability.
result First algorithm achieving optimal static and dynamic regret simultaneously against an oblivious adversary.
Study improves privacy-preserving online prediction from experts with speed-ups.
problem Privacy-preserving online prediction from experts with speed-ups.
method Differentially private federated online prediction algorithms.
result Achieves m m m -fold regret speed-up with low-loss expert in federated setting. New study shows data-oblivious attacks can outperform data-aware ones.
problem Comparing effectiveness of data-oblivious and data-aware poisoning attacks.
method Theoretical study of feature selection with LASSO, focusing on separation between full-information and oblivious attackers.
result Data-oblivious attacks can achieve the same results as full-information attacks for feature selection with LASSO.
Improved path-length regret bounds for adaptive and oblivious adversaries.
problem Adaptive and oblivious adversaries in multi-armed bandit and linear bandit problems.
method Developed two new algorithms based on optimistic mirror descent framework with novel techniques.
result Strictly improved path-length bounds for adaptive adversary and better results for oblivious adversary.
Study the tradeoffs of bandit feedback in multiclass classification.
problem The price of using bandit feedback in multiclass classification.
method Mistake bound model, analysis of variants, and comparison of learners and adversaries.
result The optimal mistake bound under bandit feedback is at most O ( k ) O(k) O ( k ) times higher than in full information, with a tight bound of O ( k ) O(k) O ( k ) . New algorithms improve online prediction from experts with privacy constraints.
problem Online prediction from experts under privacy constraints.
method Proposed and analyzed new algorithms for approximate and pure differential privacy.
result Achieved improved regret bounds for various adversaries.
New families of translation surfaces with multiple oblivious points discovered.
problem Identifying points on translation surfaces without nearby closed geodesics.
method Constructing new families of translation surfaces and proving existence in higher genera.
result Translation surfaces in every genus ≥3 have at least one oblivious point.
The paper improves smoothed analysis for online problems with adaptive adversaries.
problem Online prediction, discrepancy minimization, and online optimization with adaptive adversaries.
method General technique to prove smoothed guarantees against adaptive adversaries, reducing to simpler oblivious adversaries.
result Strong smoothed guarantees for three online problems, matching or improving previous results.
Improved upper bound for online calibrated forecasting of binary sequences.
problem Online calibrated forecasting of binary sequences.
method Introducing a variant of Qiao & Valiant's sign preservation game called sign preservation with reuse (SPR) and proving its equivalence to calibrated forecasting.
result Improved upper bound of O ( T 2 / 3 − ε ) O(T^{2/3 - \varepsilon}) O ( T 2/3 − ε ) for calibrated forecasting, improving the O ( T 2 / 3 ) O(T^{2/3}) O ( T 2/3 ) bound of Foster & Vohra. We derive upper and lower bounds for the policy regret of T T T -round online learning problems with graph-structured feedback, where the adversary is nonoblivious but assumed to have a bounded memory. We obtain upper bounds of O ~ ( T 2 / 3 ) \widetilde O(T^{2/3}) O ( T 2/3 ) and O ~ ( T 3 / 4 ) \widetilde O(T^{3/4}) O ( T 3/4 ) for strongly-observable and weakly-observab…
New algorithm recovers sparse signals robustly against Gaussian noise and adaptive adversaries.
problem Designing efficient estimators for sparse linear regression in the presence of two adversaries.
method Polynomial-time algorithms using sum-of-squares relaxations and weighted Huber loss minimization.
result Achieves error o ( ε ) o(\sqrt{\varepsilon}) o ( ε ) for various distributions and adversaries. Deep neural networks (DNNs) are powerful nonlinear architectures that are known to be robust to random perturbations of the input. However, these models are vulnerable to adversarial perturbations--small input changes crafted explicitly to fool the model. In this paper, we ask whether a DNN can distinguish adversarial …
New framework for understanding adversarial and stochastic learning.
problem Understanding the continuum from adversarial to stochastic settings in online learning.
method Distributionally constrained adversaries framework.
result Characterization of learnable distribution classes for various function classes.
We study the problem of maximizing a monotone set function subject to a cardinality constraint k k k in the setting where some number of elements τ τ τ is deleted from the returned set. The focus of this work is on the worst-case adversarial setting. While there exist constant-factor guarantees when the function is submodu…
Improved SGD for robust linear and ReLU regression with adversarial corruptions.
problem Robust regression with adversarial corruptions in streaming data.
method Stochastic gradient descent (SGD-exp) with exponentially decaying step size.
result Nearly linear convergence to true parameter with up to 50% Massart corruption rate.
Understanding and characterizing the subspaces of adversarial examples aid in studying the robustness of deep neural networks (DNNs) to adversarial perturbations. Very recently, Ma et al. (ICLR 2018) proposed to use local intrinsic dimensionality (LID) in layer-wise hidden representations of DNNs to study adversarial s…
Can machine learning models for recommendation be easily fooled? While the question has been answered for hand-engineered fake user profiles, it has not been explored for machine learned adversarial attacks. This paper attempts to close this gap. We propose a framework for generating fake user profiles which, when inco…
Paper proposes a method to detect adversarial examples that can resist norm-constrained attacks.
problem Vulnerabilities of deep neural networks to adversarial examples in sensitive domains.
method Train K binary classifiers to distinguish between clean data and adversarially perturbed samples, use at test time to classify inputs.
result Proposed method can resist norm-constrained white-box attacks.
New algorithms optimize risk-aware selection in uncertain rewards.
problem Balancing expected reward and risk in uncertain, potentially heavy-tailed rewards.
method Distribution oblivious algorithms that consider CVaR, not bound on moments/tails.
result Provable upper bounds on incorrect identification probability.
Adaptive MAB algorithms handle composite, anonymous feedback without reward interval knowledge.
problem Multi-armed bandit with composite and anonymous feedback, especially without reward interval size knowledge.
method Proposed adaptive algorithms for stochastic and adversarial cases, without reward interval knowledge.
result First algorithm for adversarial case handling non-oblivious adversary and unknown reward interval size.
Algorithm for bandits with switching costs achieves optimal regret bounds.
problem Optimal regret bounds for stochastic and adversarial bandits with switching costs.
method Adaptation of Tsallis-INF algorithm with no prior knowledge of regime or time horizon.
result Achieves minimax optimal regret bounds in various settings.
The paper tackles pure exploration in multi-armed bandits with low rank structure using oblivious sampling.
problem Pure exploration in multi-armed bandits with low rank reward sequences.
method The approach involves separating the exploration strategy from feedback, using oblivious sampling, and incorporating kernel information of reward vectors.
result Efficient algorithms with regret bound O ( d ( ln N ) / n ) O(d\sqrt{(\ln N)/n}) O ( d ( ln N ) / n ) for both time-varying and fixed cases, with a lower bound gap of O ( ln N ) O(\sqrt{\ln N}) O ( ln N ) . ABoB optimizes online configuration tuning by clustering parameters and accelerating learning.
problem Online optimization in large, dynamic parameter spaces.
method Hierarchical adversarial bandit framework.
result Significant performance gains in adversarial metric scenarios.
Algorithm adapts to shifting domains with minimal label queries.
problem Adaptive learning in online machine learning systems with domain shifts.
method Adaptive algorithm balancing regret and label queries for hidden domains.
result Achieves lower regret compared to uniform and greedy queries.
In recent years, defending adversarial perturbations to natural examples in order to build robust machine learning models trained by deep neural networks (DNNs) has become an emerging research field in the conjunction of deep learning and security. In particular, MagNet consisting of an adversary detector and a data re…
Optimal subspace embedding with near-optimal sparsity for high-dimensional data.
problem Efficiently preserving norms of vectors in high-dimensional subspaces.
method Near-optimal sparsity oblivious subspace embedding with decoupling argument and cumulant method.
result Achieved near-optimal sparsity of O ~ ( 1 / ε ) \tilde O(1/ε) O ~ ( 1/ ε ) non-zeros per column. DOFEN improves DNN performance on tabular data benchmarks.
problem DOFEN tackles the performance gap between DNNs and tree-based models on tabular data.
method DOFEN uses a two-level rODT forest ensembling process inspired by oblivious decision trees.
result DOFEN achieves state-of-the-art results on the Tabular Benchmark.
Unified framework for analyzing online convex optimization across various settings.
problem Analyzing online convex optimization in different settings and feedback types.
method Unified framework allowing systematic proposal and analysis of meta-algorithms.
result Comparable regret bounds for various feedback types and adversary types.
This work introduces oblivious fairness definitions for image generation.
problem Fairness in image generation with uncertain sensitive attributes.
method Introduces oblivious fairness definitions and uses Posterior Sampling.
result Conditional Proportional Representation can be achieved obliviously.
Paper defends against signal modulation detection attacks.
problem Minimizing intruder accuracy while maintaining receiver reliability.
method Adversarial perturbation of channel input symbols.
result Approach makes wireless communication secure with minimal performance loss.
A new federated bandit problem with multiple adversaries, solved with a near-optimal algorithm.
problem Non-stochastic federated multi-armed bandit problem with multiple adversaries.
method Proposed a near-optimal federated bandit algorithm called FEDEXP3.
result Guaranteed sub-linear regret without exchanging sequences of selected arm identities or loss sequences among agents.
New robust regression method works with fewer data points than previous methods.
problem Adversary can corrupt most of the data, making traditional regression models unreliable.
method Developed a Huber loss estimator for robust linear regression with nearly linear sample size and inverse-polynomial inlier fraction.
result The Huber loss estimator is consistent for nearly linear sample size and inverse-polynomial inlier fraction.
Algorithm learns without knowing distribution, reducing error.
problem Sequential prediction with adversarial injections and abstentions.
method Boosting procedure of weak learners for general VC classes.
result Sublinear error guarantees for general VC classes.
Develops efficient estimators for PCA and sparse regression in the presence of oblivious outliers.
problem Estimation of PCA and sparse regression in the presence of a small fraction of corrupted data.
method Designs efficient estimators using Huber loss with non-smooth regularizers like the ℓ1 norm or nuclear norm.
result Achieves consistent estimation error approaching zero as the number of observations grows.
New features generated from kernel methods are minimally dependent on sensitive features.
problem Generating fair features in the presence of sensitive and non-sensitive features.
method Relaxed Maximum Mean Discrepancy criterion, Hilbert-space-valued conditional expectation, plug-in approach.
result Closed-form solution for minimizing dependencies between new and sensitive features.
New algorithms for GLMs with oblivious noise, identifying solutions even when half the data is corrupted.
problem Regression for GLMs with additive oblivious noise.
method Distribution-independent algorithms that return accurate estimates or candidate solutions.
result First algorithmic result for GLM regression with oblivious noise, handling more than half corrupted data.
New insights into variable selection with different model assumptions.
problem Sparse recovery with ℓ ∞ \ell_\infty ℓ ∞ error guarantees in variable selection. method Separation between oblivious and adaptive models of ℓ ∞ \ell_\infty ℓ ∞ sparse recovery. result Proves a surprising contrast between oblivious and adaptive models in ℓ ∞ \ell_\infty ℓ ∞ sparse recovery. The paper develops adaptive confidence intervals for Efron's Gaussian two-groups model with unknown contamination.
problem Developing robust uncertainty quantification for Efron's Gaussian two-groups model with unknown contamination fraction.
method The approach involves Fourier-based certification procedures to find minimax-optimal adaptive confidence intervals.
result The minimax-optimal length of adaptive confidence intervals is polynomially worse than when contamination fraction is known.
A new data-oblivious sketch for logistic regression reduces data size while maintaining approximation accuracy.
problem Efficiently solving logistic regression in one pass over a data stream.
method Data-oblivious sketching approach that reduces data size to poly(μdlog n) weighted points.
result Sketching reduces data size significantly and provides approximation guarantees.
Paper improves performance guarantees for Rademacher projections.
problem Improving statistical guarantees for Rademacher random projections.
method Algebraic framework for proving Schur-concavity properties.
result Novel Schur-concavity property of Rademacher projections with improved performance.
Stochastic gradient descent on l1 loss converges to true parameter in online robust regression.
problem Online robust regression with adversarial noise.
method Stochastic gradient descent on l1 loss.
result Converges to true parameter at a rate independent of contaminated measurements.
New algorithms tackle adversarial combinatorial bandits with switching costs.
problem Adversarial combinatorial bandits with switching costs.
method Design algorithms operating in batches to restrict switches, proving lower bounds and achieving upper bounds on regret.
result Achieved upper bounds on regret for both bandit and semi-bandit feedback settings.
Improves bandits with knapsacks guarantees for partially stochastic workloads.
problem Improves guarantees for Bandits with Knapsacks (BwK) with partially stochastic workloads.
method Defines Approximately Stationary BwK, explores algorithms with smooth competitive ratios transitioning between stochastic and adversarial cases.
result Offers competitive ratios that smoothly transition between the best possible guarantees in stochastic and adversarial cases, especially beneficial when budget is small.
Study shows efficient algorithms for noiseless linear regression require quadratic sample complexity in contamination rate.
problem Efficient algorithms for noiseless linear regression under Gaussian covariates with oblivious contamination.
method Formal evidence using Statistical Query complexity.
result Any efficient Statistical Query algorithm requires VSTAT complexity at least Ω(d^(1/2)/α^2).
New method for robust regression with near-optimal performance even with high corruption rates.
problem Robust linear regression with response variable corruptions.
method Adaptive hard thresholding for consistent estimation.
result Near-optimal consistent estimation of the true regression vector with 1 − o ( 1 ) 1-o(1) 1 − o ( 1 ) fraction of corruptions. New algorithms optimize without knowing problem parameters.
problem Optimizing large-scale problems without knowing key parameters.
method Combining mirror descent with dual averaging techniques.
result Converges without prior knowledge of problem parameters.
NODE improves deep learning on tabular data, outperforming GBDT.
problem Limited performance of deep learning on tabular data compared to gradient boosting decision trees.
method Introducing Neural Oblivious Decision Ensembles (NODE), a deep learning architecture that generalizes ensembles of oblivious decision trees.
result NODE outperforms leading GBDT packages on most tabular tasks.