The study examines how verifier imperfections impact test-time scaling techniques.
problem Understanding how verifier imperfections affect test-time scaling methods.
method Proves the instance-level accuracy of Best-of-N and Rejection Sampling methods using the geometry of the verifier's ROC curve.
result RS outperforms BoN for fixed compute, but both converge to the same accuracy in the infinite-compute limit.
Efficiently verifies neural networks by handling neuron splits, improving speed and accuracy.
problem Handling neuron split constraints in incomplete neural network verification.
method β-CROWN, which optimizes parameters β to encode neuron splits and uses them in bound propagation.
result β-CROWN significantly speeds up verification while maintaining high accuracy.
MixTrain improves verifiable robustness of neural networks without sacrificing efficiency.
problem Efficiently making neural networks robust against adversarial attacks.
method Stochastic robust approximation and dynamic mixed training techniques.
result Achieves up to 95.2% verified robust accuracy with significantly reduced training time.
New method preserves both benign and robust accuracy in deep neural networks.
problem Large neural networks have high computational and storage costs.
method Pruning with fine-tuning to preserve both benign and robust accuracy.
result Average 93% benign accuracy, 92.5% empirical robust accuracy, and 85.0% verifiable robust accuracy preserved while compressing by 10x.
New expressive losses improve adversarial robustness without sacrificing accuracy.
problem Training networks for robustness at the expense of accuracy.
method Formalizing expressivity, using convex combinations of adversarial attacks and IBP bounds.
result Trivial expressive losses yield state-of-the-art results in various settings.
This work improves neural network robustness to symbol substitutions using formal verification.
problem Neural networks' vulnerability to adversarial attacks, especially under discrete text perturbations.
method Formal verification using Interval Bound Propagation on a simplex model of input perturbations.
result Models show improved verified accuracy under perturbations with formal guarantees.
IBP-R improves verified adversarial robustness with simple, effective interval bound propagation.
problem Improving verifiability of adversarially trained networks.
method Coupling adversarial attacks with interval bound propagation for minimized verification gap.
result State-of-the-art verified robustness-accuracy trade-offs for small perturbations on CIFAR-10.
CROP verifies clean prefixes in reasoning traces, improving downstream repair accuracy.
problem Uncertainty in reasoning traces prevents full certification of entire responses.
method CROP selects a calibrated threshold to certify the longest prefix with low risk proxies.
result CROP improves downstream repair accuracy by preserving valid reasoning and discarding misleading suffixes.
Unified understanding of neural networks on group operations verified.
problem Understanding and verifying neural networks trained on group operations.
method Investigated one-hidden-layer neural networks trained on binary operation of finite groups, revealing structure and providing a compact proof of model performance.
result Verified explanation applies to a large fraction of networks trained on the symmetric group S5, providing a >=95% accuracy bound for 45% of models.
IBP simplifies robust training for large networks.
problem Training robust neural networks with scalable methods.
method Interval Bound Propagation (IBP) for robust training.
result IBP enables training large provably robust networks.
New approach verifies tree ensembles efficiently, making them more robust.
problem Verifying robustness of decision tree ensembles against evasion attacks is NP-hard.
method Identified large-spread ensembles, proposed verifiable learning, designed training algorithm.
result Large-spread ensembles can be verified in polynomial time and are more robust.
Paper proposes dense average network for improved power load forecasting.
problem Improving power load forecasting accuracy to save millions for the power industry.
method Introduces dense average connection and constructs dense average network for power load forecasting.
result Proposed model outperforms existing methods on public datasets.
Method verifies robustness of graph neural networks to graph perturbations.
problem Vulnerability of graph neural networks to adversarial attacks.
method Proposes efficient method for certifying robustness to graph perturbations.
result Efficient computation of robustness certificates for graph neural networks.
We show how to compute the Bayes error-rate for speaker verifiers.
problem How many errors does a speaker verifier make in a hundred trials?
method We compute the Bayes error-rate using calibrated likelihood ratios and user-supplied prior probabilities.
result The Bayes error-rate is upper bounded by the minimum of EER, P, and 1-P.
New approach uses interpolation models and error bounds for verifiable scientific machine learning.
problem Challenges in verifying and validating modern scientific machine learning workflows.
method Combines multiple standard interpolation techniques with error bounds for efficient computation and comparative performance analysis.
result Error bounds for interpolation techniques can be computed or estimated efficiently, aiding in validation goals.
VERAFI improves financial AI by verifying calculations and compliance.
problem Financial AI systems generate errors and violations during reasoning.
method VERAFI combines dense retrieval, reranking, and automated reasoning policies.
result VERAFI achieves 94.7% factual correctness, 81% relative improvement.
We introduce a general framework to handle structured models (sparse and block-sparse with possibly overlapping blocks). We discuss new methods for their recovery from incomplete observation, corrupted with deterministic and stochastic noise, using block-ℓ1 regularization. While the current theory provides promis…
Linear trends in classifier accuracy observed under distribution shift.
problem Understanding why classifier accuracies show linear trends under distribution shift.
method Assumed model similarity and verified empirically.
result Linear trend in classifier accuracy occurs unless distribution shift is large.
EORM boosts LLM accuracy with a lightweight, energy-based verifier.
problem Efficiently verifying mathematical reasoning in large language models.
method Energy-based framework to rank Chain-of-Thought solutions using simple outcome labels.
result EORM boosts LLM accuracy to 90.7% on GSM8k and 63.7% on MATH with only 55M parameters.
Data augmentation methods improve worst-case model performance.
problem Ensuring fair predictions across subpopulations in large models.
method Linear last layer retraining with data augmentation techniques.
result Optimal worst-group accuracy achieved for Gaussian latent representation distribution.
Combines neural networks and logic circuits for interpretable, accurate, and cost-effective learning.
problem Lack of generalizability and interpretability in neural networks and high hardware cost in logic circuits.
method Trains a neural network, then translates it to random forests, and finally to AND-Inverter logic.
result The pipeline maintains greater accuracy and minimizes logic complexity.
Neural network accuracy improves with denser training samples.
problem Improving neural network accuracy on unseen test samples.
method Bounding empirical training error smoothed across activation regions and using it to discard high-risk test samples.
result Discarding high-risk test samples based on error bounds improves prediction accuracy by up to 20%.
Unified approach improves accuracy in private estimation.
problem Protecting sensitive information in statistical estimation.
method Unified Bayes Private Point (UBaPP) approach to Bayes point estimation under differential privacy constraints.
result Achieves better accuracy-privacy trade-off than traditional methods.
Analytical pricing formulas and Greeks are obtained for European and American basket put options using Mellin transforms. We assume assets are driven by geometric Brownian motion which exhibit correlation and pay a continuous dividend rate. A novel approach to numerical Mellin inversion is achieved via the fast Fourier…
New method certifies global robustness of neural networks efficiently.
problem Adversarial examples threaten certifiably robust neural networks.
method Formalized global robustness, adapted widely-used architectures with efficient global Lipschitz bounds.
result Certifiable robust models achieve state-of-the-art verifiable accuracy with negligible costs.
In this study, we introduce an explicit trading-volume process into the Almgren-Chriss model, which is a standard model for optimal execution. We propose a penalization method for deriving a verification theorem for an adaptive optimization problem. We also discuss the optimality of the volume-weighted average-price st…
The study evaluates different probability models for uncertainty visualization using entropy calculations.
problem Choosing the right probability model affects memory use, run time, and accuracy in uncertainty visualization.
method Entropy calculation on ensemble data to compare various probability models (uniform, Gaussian, histogram, quantile).
result Models matching the ensemble data distribution have the lowest entropy, indicating better accuracy.
New bootstraps improve speed and accuracy for graph count functionals.
problem Efficiently counting subgraphs in large graphs.
method Developed two types of multiplier bootstraps: a fast, approximate linear one and a quadratic one for denser graphs.
result Both bootstraps provide valid inference and higher-order accuracy under different graph sparsity conditions.
A novel ν-SVQR model for automatic quantile estimation.
problem Quantile estimation with automatic accuracy control.
method Uses ν fraction of training data points and asymmetric ε-insensitive zone. result Automatic control over accuracy achieved through ν-SVQR model. This paper analyzes the trade-off between accuracy and communication in personalized federated learning.
problem The accuracy-communication trade-off in personalized federated learning.
method The paper provides a quantitative characterization of the personalization degree on the trade-off, establishing minimax optimality.
result The paper offers theoretical insights for choosing the personalization degree and validates the results on synthetic and real-world datasets.
Deep models maximize minimum margin for high accuracy but decrease average margin, leading to poor robustness.
problem Inadequate balance between accuracy and robustness in deep model training.
method Analyzed the training process of deep models and proposed a new regularizer to promote average margin.
result Demonstrated an intrinsic trade-off between accuracy and robustness, and proposed a regularizer to improve robustness.
New methods improve statistical accuracy of complex models without high computational cost.
problem Improving statistical accuracy of complex models without high computational cost.
method Neural posterior and likelihood estimation (NPE and NLE) methods.
result NPE and NLE methods have similar theoretical guarantees to ABC and BSL, but achieve accuracy at a reduced computational cost.
Paper proposes verifier engineering for improving foundation models.
problem Challenges in providing effective supervision signals for foundation models.
method Leverages automated verifiers to perform verification tasks and deliver feedback.
result Verifier engineering can enhance foundation models' capabilities.
Deep neural nets classify sleep stages from raw PSG data with high accuracy.
problem Automatically classifying sleep stages from raw polysomnogram signals.
method Deep residual neural networks trained on 50 convolutional layers.
result Best model achieved 84.1% accuracy and 0.746 Cohen's kappa.
A hybrid model reduces graph complexity for improved classification accuracy.
problem High computational complexity and large number of parameters in higher-order graph convolutional networks.
method Weight sharing mechanism and novel fusion pooling layer to reduce parameters and complexity.
result The proposed model achieves highest classification accuracy with fewer trainable parameters.
Validates neural networks inputs to protect against adversarial examples.
problem Ensuring neural networks robustness against adversarial attacks.
method Runtime local robustness verification based on normal distribution of robustness radii.
result Improves neural network accuracy and protects against adversarial examples.
This paper proposes a new algorithmic framework, predictor-verifier training, to train neural networks that are verifiable, i.e., networks that provably satisfy some desired input-output properties. The key idea is to simultaneously train two networks: a predictor network that performs the task at hand,e.g., predicting…
We present a novel modulation level classification (MLC) method based on probability distribution distance functions. The proposed method uses modified Kuiper and Kolmogorov-Smirnov distances to achieve low computational complexity and outperforms the state of the art methods based on cumulants and goodness-of-fit test…
SGD with large learning rates can achieve better test accuracy than expected.
problem SGD with large learning rates often outperforms expected convergence bounds.
method Proved that SGD with small learning rates stays close to gradient flow path on modified loss.
result Explicitly adding an implicit regularizer to the loss improves test accuracy.
In this paper, we firstly introduce a method to efficiently implement large-scale high-dimensional convolution with realistic memristor-based circuit components. An experiment verified simulator is adapted for accurate prediction of analog crossbar behavior. An improved conversion algorithm is developed to convert conv…
An ensemble of randomized NNs improves time series forecasting accuracy.
problem Forecasting time series with multiple seasonality and nonstationarity.
method Randomized neural networks with pattern-based time series representation and diversity control strategies.
result Outperforms statistical and machine learning models in forecasting accuracy.
SG-NTF completes HDI tensors with spectral mapping and spatio-temporal gating.
problem High-dimensional and incomplete tensor completion.
method Spectra-Guided Neural Tucker Factorization (SG-NTF) with Spatio-Temporal Co-Gating (STCG).
result Maintains competitive completion accuracy with parameter efficiency.
We consider the problem of designing models to leverage a recently introduced approximate model averaging technique called dropout. We define a simple new model called maxout (so named because its output is the max of a set of inputs, and because it is a natural companion to dropout) designed to both facilitate optimiz…
Study on TVL computation in DeFi protocols, proposing verifiable metrics.
problem Lack of standardization and verifiability in TVL computation.
method Systematic study of 939 DeFi projects, analyzing methodologies and proposing vTVL.
result 240 protocols use repeated balance queries, limiting verifiability.
The paper explores the incompatibility between fair privacy, need-to-know, and fairness in classifier outputs.
problem The interaction between fair privacy, need-to-know, and fairness in classifier outputs.
method Formulated and explored the interaction between fair privacy, need-to-know, and fairness in classifier outputs.
result Optimal classifiers are generally incompatible with fair privacy and need-to-know.
In recent years an increasing number of researchers and practitioners have been suggesting algorithms for large-scale neural network architecture search: genetic algorithms, reinforcement learning, learning curve extrapolation, and accuracy predictors. None of them, however, demonstrated high-performance without traini…
New regularization techniques improve stability of deep neural networks.
problem Improving stability of deep neural networks in high-dimensional data.
method Apply manifold regularization to develop new regularizers based on graph Laplacian sparsification.
result Empirically, networks achieve high stability in various perturbation models, including adversarial attacks.
SwishNet improves speech, music, and noise classification and segmentation.
problem Speech, Music, and Noise classification/segmentation for audio processing.
method Proposes a fast and lightweight 1D CNN (SwishNet) for MFCC features, trained with knowledge distillation from ImageNet.
result Achieved high accuracy (>97% clip classification, >93% frame-wise segmentation) on MUSAN corpus.