Paper introduces new methods for modeling categorical data.
problem Training generative models on categorical data like text and segmentation.
method Argmax Flows and Multinomial Diffusion models.
result Models outperform existing methods in log-likelihood.
New algorithm explores optimally using same primitives as Thompson Sampling.
problem Optimal exploration of unknown parameter vectors with noisy measurements.
method Algorithm leveraging sampling and argmax oracles.
result Achieves exponential convergence rate, optimal among all allocations.
Proposes a stable classifier using inflated argmax for multiclass classification.
problem Inherent instability of taking the maximizer in multiclass classification.
method Bagging for stable continuous scores, inflated argmax for stable labels.
result Inflated argmax provides necessary protection against unstable classifiers without loss of accuracy.
Investigates statistical properties of perturb-softmax and perturb-argmax distributions.
problem Underexplored statistical properties of Gumbel-Softmax and Gumbel-Argmax distributions.
method Investigates convexity and differentiability to determine completeness and minimality of these distributions.
result Identifies parameters that admit complete and minimal representation of probability distributions.
New algorithm reduces contextual bandit identification to argmax calls.
problem Best-arm identification in stochastic contextual bandits.
method Instance-optimal PAC algorithm using argmax oracle calls.
result First instance-dependent PAC sample complexity for contextual bandits.
Study improves estimation of rare language model outputs.
problem Estimating probabilities of rare outputs in language models.
method Importance sampling vs. activation extrapolation for low probability estimation.
result Importance sampling outperforms activation extrapolation.
Introduces Fitzpatrick losses, tighter than Fenchel-Young losses.
problem Improving loss functions for machine learning.
method Introduces Fitzpatrick losses based on the Fitzpatrick function.
result Fitzpatrick losses are tighter than Fenchel-Young losses.
New method stabilizes model selection with theoretical guarantees.
problem Stability of model selection methods in the face of noisy or incomplete data.
method Combines bagging with an 'inflated' argmax operation to select a stable set of models.
result Stable model selection with high probability of overlap after removing any data point.
Paper proposes new loss functions for training energy networks.
problem Challenges in computing gradients for training energy networks.
method Proposes generalized Fenchel-Young losses for efficient gradient computation.
result Demonstrates the calibration of excess risk for linear-concave energies.
Efficient algorithms for planning in cooperative multi-agent reinforcement learning with combinatorial action spaces.
problem Planning in cooperative multi-agent reinforcement learning with a combinatorial action space.
method Efficient algorithms using local access to a simulator and linear function approximation, with improvements for additive feature decomposition and kernelized settings.
result Polynomial compute and query complexity in relevant problem parameters.
ZOSPI improves RL policies with global value function exploitation.
problem Limited sample efficiency of PG algorithms in RL.
method Zeroth-order supervised policy improvement, leveraging global value function estimation.
result ZOSPI achieves competitive results with remarkable sample efficiency.
The top-k error is often employed to evaluate performance for challenging classification tasks in computer vision as it is designed to compensate for ambiguity in ground truth labels. This practical success motivates our theoretical analysis of consistent top-k classification. Surprisingly, it is not rigorously und…
We propose a novel neural sequence prediction method based on \textit{error-correcting output codes} that avoids exact softmax normalization and allows for a tradeoff between speed and performance. Instead of minimizing measures between the predicted probability distribution and true distribution, we use error-correcti…
Estimates change point in high-dimensional dynamic graphical models.
problem Detecting change points in high-dimensional graphical models.
method Developed an estimator with Op(ψ−2) rate of convergence, established asymptotic distribution under high-dimensional scaling. result Asymptotic distribution characterized under vanishing and non-vanishing jump size regimes.
New algorithm identifies optimal actions in large reward spaces efficiently.
problem Finding the best action from a large set of options with minimal trials.
method GenTS-Explore algorithm for real-valued combinatorial pure exploration.
result Achieves optimal sample complexity for large action sets.
New bounds show transformers need longer training for length generalization.
problem Understanding when transformers can generalize to longer inputs.
method Analyzing different settings of transformers, providing quantitative bounds.
result Transformers need training data longer than previously thought for length generalization.
Attention mechanisms and non-local mean operations in general are key ingredients in many state-of-the-art deep learning techniques. In particular, the Transformer model based on multi-head self-attention has recently achieved great success in natural language processing and computer vision. However, the vanilla algori…
These notes aim to shed light on the recently proposed structured projected intermediate gradient optimization technique (SPIGOT, Peng et al., 2018). SPIGOT is a variant of the straight-through estimator (Bengio et al., 2013) which bypasses gradients of the argmax function by back-propagating a surrogate "gradient." We…
The most direct way to express arbitrary dependencies in datasets is to estimate the joint distribution and to apply afterwards the argmax-function to obtain the mode of the corresponding conditional distribution. This method is in practice difficult, because it requires a global optimization of a complicated function,…
The paper refines and generalizes worst-case law invariant convex risk measures.
problem Developing robust convex risk measures under uncertainty sets.
method Generalizing closed forms for worst-case law invariant convex risk measures with uncertainty sets based on norms and moment constraints.
result Explicit closed forms for convex risk measures are developed and assessed through numerical simulations.
RankSEG-RMA improves semantic segmentation efficiency and applicability.
problem Inconsistent or suboptimal semantic segmentation results due to argmax or thresholding.
method Developed RankSEG-RMA using reciprocal moment approximation to optimize Dice and IoU metrics.
result RankSEG-RMA reduces computational complexity to O(d) while maintaining comparable performance.
Protocol diagnoses neural HJB-PIDE solvers for Lévy jumps, revealing a missing factor in their importance-proposal density.
problem Neural PDE solvers can match scalar diagnostics but miscompute operators, leading to systematic errors.
method Five-step diagnostic protocol decomposes neural solve into components, compares them with independent reference solutions.
result Corrected a missing 1/2-mixture factor in the neural method's importance-proposal density, improving control accuracy.
Develops a novel ML smoothing method for incomplete data in state-space models.
problem Estimating states in stochastic systems with incomplete information.
method Introduces score function and conditional observed information matrices for incomplete data, and uses them to derive the ML smoother.
result The ML smoother provides more accurate state estimates with lower standard errors compared to the standard ML state estimator.
Recently, there has been rising interest in Bayesian optimization -- the optimization of an unknown function with assumptions usually expressed by a Gaussian Process (GP) prior. We study an optimization strategy that directly uses an estimate of the argmax of the function. This strategy offers both practical and theore…
Entropy regularization improves policy optimization in reinforcement learning.
problem Improving policy optimization in reinforcement learning.
method Entropy regularization is introduced to soften the greedy policy towards a more diverse softmax policy, leading to a continuously parameterized algorithm that interpolates between policy gradient and Q-learning.
result An intermediate algorithm can improve performance in reinforcement learning.
We consider the problem of n-class classification (n≥2), where the classifier can choose to abstain from making predictions at a given cost, say, a factor α of the cost of misclassification. Designing consistent algorithms for such n-class classification problems with a `reject option' is the main goal of t…
Adapts pretrained models to new classes without additional training.
problem Training models on limited labels and predicting new, unseen classes.
method Uses Fréchet mean instead of argmax for prediction, leveraging metric space distances.
result Improves model performance on unseen classes by up to 29.7% on ImageNet.
Estimates change point in high dimensional time series models.
problem Change point estimation in high dimensional time series.
method Plug-in least squares estimator with sufficient conditions for adaptivity.
result Optimal rate of convergence Op(ξ−2) in integer scale. Introduces a differentiable approximation to the zero-one loss.
problem Incompatibility of zero-one loss with gradient-based optimization.
method Smooth projection onto hypersimplex through constrained optimization.
result Achieves significant improvements in generalization under large-batch training.
There are three quite distinct ways to train a machine learning model on recommender system logs. The first method is to model the reward prediction for each possible recommendation to the user, at the scoring time the best recommendation is found by computing an argmax over the personalized recommendations. This metho…
There is a fast-growing literature on estimating optimal treatment regimes based on randomized trials or observational studies under a key identifying condition of no unmeasured confounding. Because confounding by unmeasured factors cannot generally be ruled out with certainty in observational studies or randomized tri…
Structured prediction energy networks (SPENs; Belanger & McCallum 2016) use neural network architectures to define energy functions that can capture arbitrary dependencies among parts of structured outputs. Prior work used gradient descent for inference, relaxing the structured output to a set of continuous variables a…
The marginal maximum a posteriori probability (MAP) estimation problem, which calculates the mode of the marginal posterior distribution of a subset of variables with the remaining variables marginalized, is an important inference problem in many models, such as those with hidden variables or uncertain parameters. Unfo…
In this paper we introduce the transductive linear bandit problem: given a set of measurement vectors X⊂Rd, a set of items Z⊂Rd, a fixed confidence δ, and an unknown vector θ∗∈Rd, the goal is to infer $\text{argmax}_{z\in \mathcal{Z}} z^\t…
In this dissertation, we focus on several important problems in structured prediction. In structured prediction, the label has a rich intrinsic substructure, and the loss varies with respect to the predicted label and the true label pair. Structured SVM is an extension of binary SVM to adapt to such structured tasks. I…
Estimates sub-Gaussian parameter with consistent and optimal rates.
problem Estimating sub-Gaussian parameter from random variables.
method Constrained maximization of empirical weighted cumulant generating function.
result Root-n rate estimator is consistent and optimal under certain conditions.
Enhanced Sampling Scheme improves masked generative modeling.
problem Limitations of existing sampling schemes in masked non-autoregressive generative modeling.
method ESS consists of three stages: Naive Iterative Decoding, Critical Reverse Sampling, and Critical Resampling.
result ESS achieves significant performance gains in unconditional and class-conditional sampling.
PRS improves rejection sampling by learning better proposals.
problem High rejection rate in traditional rejection sampling.
method PRS uses a kernel estimator to learn better sampling proposals.
result PRS guarantees a low number of accepted samples.
This paper reviews various sampling methods from statistics and machine learning.
problem Addressing sampling methods in statistics and machine learning.
method Explains and reviews simple random sampling, bootstrapping, stratified sampling, cluster sampling, multistage sampling, network sampling, snowball sampling, and sampling from cumulative distribution function.
result Summarizes characteristics, pros, and cons of different sampling methods.
RISA improves VFL by using imputed samples with low uncertainty.
problem Limited overlapping samples constrain VFL performance.
method Imputing non-overlapping samples and using evidence theory to select reliable imputed samples.
result Significant performance gains achieved, especially with limited overlapping samples.
Improved privacy-preserving methods for estimating multiple samples from distributions.
problem Estimating multiple samples from distributions while maintaining privacy.
method Developed new multi-sampling techniques for differentially private data estimation.
result Achieved significant reduction in sample complexity for multi-sampling from finite domains and Gaussian distributions.
Paper introduces a new sampling method combining Consistency Models with importance sampling.
problem Inherent errors in samples and high NFEs for high-quality samples in Boltzmann distributions.
method Combines Consistency Models with importance sampling to produce unbiased samples with minimal NFEs.
result Produces unbiased samples using only 6-25 NFEs, comparable to 100 NFEs for DDPMs.
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%.
Wedge Sampling improves tensor completion with nearly-linear sample complexity.
problem Efficiently completing low-rank tensors from a subset of entries.
method Non-adaptive wedge sampling to promote structured connections in tensor completion.
result Polynomial-time algorithms achieve weak and exact recovery with nearly linear sample complexity.
Algorithm samples constrained stochastic differential equations.
problem Sampling stochastic differential equations with complex constraints.
method Pathspace Metropolis-adjusted manifold sampling.
result Demonstrated effectiveness in various constrained conditions.
In modern data analysis, random sampling is an efficient and widely-used strategy to overcome the computational difficulties brought by large sample size. In previous studies, researchers conducted random sampling which is according to the input data but independent on the response variable, however the response variab…
Generative models map simple samples to complex target samples.
problem Improving Monte-Carlo sampling techniques.
method Variational learning of dynamical maps between base and target measures.
result Improved sampling efficiency through feedback loops.
Sampling is a fundamental problem in computer science and statistics. However, for a given task and stream, it is often not possible to choose good sampling probabilities in advance. We derive a general framework for adaptively changing the sampling probabilities via a collection of thresholds.In general, adaptive samp…