Improves text-to-image diffusion models using GFlowNets.
problem Aligning diffusion models with text descriptions.
method Post-training diffusion models with GFlowNets to generate high-reward images.
result Effective alignment of large-scale text-to-image diffusion models with reward information.
Enhances reward specification in RL with a novel language-based approach.
problem Reward specification in RL can lead to unintended, potentially harmful behaviours.
method Developed a novel class of language-based Reward Machines using RML's built-in memory.
result Can specify non-regular, non-Markovian reward functions for complex tasks.
A new approach for specifying and synthesizing subroutines for optimizing metrics.
problem Specifying and optimizing subroutines for various metrics.
method Formalizing programming by rewards (PBR), using continuous-optimization techniques to synthesize decision functions as if-then-else programs.
result Synthesized decision functions are optimal in cases when rewards have nice properties.
Paper tackles constrained bandit problems with a new learning framework.
problem Optimizing a black-box reward function subject to a black-box constraint function over a continuous space.
method Rectified Pessimistic-Optimistic Learning (RPOL) framework, incorporating optimistic and pessimistic GP bandit learning.
result RPOL achieves sublinear regret and minimal cumulative constraint violation.
A new framework isolates exploration challenges in RL without explicit rewards.
problem Challenges in reinforcement learning, especially exploration.
method Reward-free RL framework, collecting trajectories without a reward function, then computing policies for various reward functions.
result Efficient algorithm that conducts exploration and computes near-optimal policies for multiple reward functions.
LIMIS improves locally interpretable models by selecting and distilling key instances.
problem Low fidelity of locally interpretable models.
method LIMIS uses instance-wise subsampling guided by policy gradient and reward to improve fidelity.
result LIMIS near-matches black-box model accuracy while significantly improving fidelity.
Partially observable Markov decision processes (POMDPs) are a powerful abstraction for tasks that require decision making under uncertainty, and capture a wide range of real world tasks. Today, effective planning approaches exist that generate effective strategies given black-box models of a POMDP task. Yet, an open qu…
We explore the use of Evolution Strategies (ES), a class of black box optimization algorithms, as an alternative to popular MDP-based RL techniques such as Q-learning and Policy Gradients. Experiments on MuJoCo and Atari show that ES is a viable solution strategy that scales extremely well with the number of CPUs avail…
New model-free RL algorithm tackles robust average-reward problems with finite sample complexity analysis.
problem Long-term decision-making in environments with varying dynamics.
method Proposes Robust Halpern Iteration (RHI) algorithm based on a black-box sampling oracle and multi-level Monte-Carlo estimator.
result Achieves ε-optimal robust policy with sample complexity of O(1/ε^(2+o(1))) under generative model setting.
Master algorithm selects best contextual bandit from a collection.
problem Model selection in stochastic contextual bandit setting.
method Random selection with probability adjustment based on comparison of cumulative rewards.
result Achieves the same regret rate as the best candidate in a collection of black-box algorithms.
Signed compression progress on a sealed audit is goodhart-resistant.
problem Intrinsic motivation for agents to improve their world models by compressing experience.
method Rewarding agents for the signed decrease of a fixed sealed-audit loss.
result Cumulative reward telescopes exactly to endpoint audit improvement, preventing infinite reward push while true audit performance stagnates.
This work explores adaptations of successful multi-armed bandits policies to the online contextual bandits scenario with binary rewards using binary classification algorithms such as logistic regression as black-box oracles. Some of these adaptations are achieved through bootstrapping or approximate bootstrapping, whil…
New algorithms for efficient return distribution approximation in reinforcement learning.
problem Efficiently approximating unknown return distributions in reinforcement learning.
method Introduced novel distributional dynamic programming algorithms for arbitrary probabilistic reward mechanisms.
result Proved error bounds for the algorithms in Wasserstein and Kolmogorov--Smirnov distances.
POETS optimizes LLMs by combining policy ensembles and KL regularization.
problem Balancing exploration and exploitation in decision-making and optimization.
method POETS uses policy ensembles and KL regularization to optimize LLMs efficiently.
result POETS achieves state-of-the-art sample efficiency across various domains.
Study validates metrics for offline MBO using diffusion models.
problem Evaluate metrics for offline MBO without ground truth oracle.
method Propose and quantify validation metrics over datasets.
result Identify most effective validation metrics.
Improves inference-time alignment for diffusion models without updating weights.
problem Aligning diffusion models without updating weights for high-reward outputs.
method Trust-Region Iterative Twisted Sequential Monte Carlo (TRI-TSMC) for variance reduction and efficiency.
result Improves primary alignment objectives on text generation tasks.
Bayesian optimization under heavy-tailed noise with improved regret bounds.
problem Optimizing functions with heavy-tailed noise.
method Adaptation of GP-UCB with reward truncation and novel kernel approximation techniques.
result Regret bounds matching the lower bound for SE kernel, improving upon existing methods.
Managers of US National Forests must decide what policy to apply for dealing with lightning-caused wildfires. Conflicts among stakeholders (e.g., timber companies, home owners, and wildlife biologists) have often led to spirited political debates and even violent eco-terrorism. One way to transform these conflicts into…
Deep learning models are vulnerable to external attacks. In this paper, we propose a Reinforcement Learning (RL) based approach to generate adversarial examples for the pre-trained (target) models. We assume a semi black-box setting where the only access an adversary has to the target model is the class probabilities o…
Spanning attack improves black-box attacks with unlabeled data.
problem Query inefficiency in black-box attacks due to high input space dimensionality.
method Proposes spanning attack by constraining adversarial perturbations in a low-dimensional subspace via an auxiliary unlabeled dataset.
result Significantly improves query efficiency of black-box attacks.
Unified approach for sequence design combining likelihood-free inference and black-box optimization.
problem Designing biological sequences efficiently and accurately.
method Unified probabilistic framework integrating likelihood-free inference and black-box optimization.
result Previous optimization methods can be adapted and new algorithms proposed within this framework.
Black box machine learning models are currently being used for high stakes decision-making throughout society, causing problems throughout healthcare, criminal justice, and in other domains. People have hoped that creating methods for explaining these black box models will alleviate some of these problems, but trying t…
Interpretable companion model for black-box classifiers.
problem Dilemma between interpretable and black-box models.
method Trains a companion model from data and black-box model predictions, optimizing a combination of accuracy and complexity.
result Companion model provides interpretable predictions with a slight accuracy loss for user choice.
Understanding how a learned black box works is of crucial interest for the future of Machine Learning. In this paper, we pioneer the question of the global interpretability of learned black box models that assign numerical values to symbolic sequential data. To tackle that task, we propose a spectral algorithm for the …
Control policies, trained using the Deep Reinforcement Learning, have been recently shown to be vulnerable to adversarial attacks introducing even very small perturbations to the policy input. The attacks proposed so far have been designed using heuristics, and build on existing adversarial example crafting techniques …
Black-box risk scoring models permeate our lives, yet are typically proprietary or opaque. We propose Distill-and-Compare, a model distillation and comparison approach to audit such models. To gain insight into black-box models, we treat them as teachers, training transparent student models to mimic the risk scores ass…
New research shows many recent defenses against adversarial examples are ineffective against black-box attacks.
problem The robustness of recent defenses against adversarial examples is insufficient, especially against black-box attacks.
method Evaluation of nine defenses on two black-box adversarial models and six attacks on CIFAR-10 and Fashion-MNIST datasets.
result Most recent defenses provide only marginal improvements in security (<25%) compared to undefended networks. New method optimizes black-box functions using generative models and Wasserstein distance.
problem Optimizing black-box functions with stochastic responses in high dimensions.
method Deep generative surrogate models and Wasserstein distance for uncertainty estimation.
result Method outperforms state-of-the-art methods in robustness to function shape and stochasticity.
Paper proposes a new black-box attack approach to minimize visual distortion.
problem Constructing adversarial examples that minimize visual distortion in a black-box threat model.
method Learning the noise distribution of adversarial examples to approximate the gradient of a non-differentiable loss function.
result The proposed attack results in much lower visual distortion compared to state-of-the-art black-box attacks.
REST improves robustness of black-box models to geometric transformations.
problem Overconfident incorrect predictions on out-of-distribution samples.
method REinforcement Spatial Transform learner (REST) that transforms input data into in-distribution samples.
result Improves robustness to geometric transformations and sample efficiency.
Adapts model-based advice to stabilize black-box policies for nonlinear control.
problem Stabilizing machine-learned policies for nonlinear control with limited model information.
method Proposes an adaptive λ-confident policy to combine black-box and model-based advice. result Proves the stability of the adaptive λ-confident policy and its competitive ratio. Deep neural networks work well at approximating complicated functions when provided with data and trained by gradient descent methods. At the same time, there is a vast amount of existing functions that programmatically solve different tasks in a precise manner eliminating the need for training. In many cases, it is po…
MALC combines interpretable linear models with black-box models for better predictions and transparency.
problem Combining interpretability with black-box models for better predictions.
method Formulates MALC as a convex optimization problem and uses accelerated proximal gradient method for training.
result MALC provides an efficient frontier balancing prediction accuracy and transparency.
NP-Attack reduces query counts for black-box adversarial attacks.
problem Efficiency of black-box adversarial attacks is low.
method Uses Neural Process to characterize image structure and find adversarial examples.
result NP-Attack significantly decreases query counts.
Optimizes ad pruning in sponsored search systems using reinforcement learning.
problem How to efficiently select top K ads from N candidates to maximize revenue.
method Model-free reinforcement learning approach considering downstream as a black-box environment.
result Remarkable improvements in revenue achieved through reinforcement learning.
A new adversarial attack method using structured search and contextual bandits.
problem Black-box adversarial attacks on deep learning models.
method Structured search space and Bayesian optimization for contextual bandits.
result Achieves state-of-the-art success rates and query efficiencies.
Bayesian optimisation generates saliency maps for black-box models.
problem Generating saliency maps for models without access to parameters.
method Bayesian optimisation sampling method to find global salient regions.
result Approach outperforms grid-based methods and performs similarly to gradient-based methods.
New approach improves black-box planning efficiency by discovering focused macros.
problem Difficulty of deterministic planning increases exponentially with depth.
method Discovering macro-actions with focused effects to improve goal-count heuristics.
result Focused macros dramatically improve black-box planning efficiency.
Reward hacking exploits misspecified rewards, affecting agent capabilities and true performance.
problem Reward hacking in RL models exploiting reward misspecifications.
method Constructed four RL environments with misspecified rewards; analyzed agent capabilities and behavior.
result More capable agents exploit reward misspecifications, achieving higher proxy reward but lower true reward.
New method improves black-box attacks using pre-trained models.
problem Efficiently attacking black-box models with limited information.
method EigenBA algorithm leveraging pre-trained white-box model's Jacobian matrix.
result Optimal perturbations are related to right singular vectors of Jacobian matrix.
A RL-enhanced quantum-inspired algorithm solves combinatorial optimization problems.
problem Optimizing quantum-inspired algorithms for combinatorial problems.
method Reinforcement learning agent tunes hyperparameters of a quantum-inspired algorithm.
result The RL-enhanced algorithm samples high-quality solutions to the Ising problem.
Efficient method for generating adversarial examples with limited query budget.
problem Developing black-box adversarial attacks with limited information.
method Bayesian Optimization in a structured low-dimensional subspace.
result Significantly higher attack success rate with fewer queries.
Interpretable machine learning has become a strong competitor for traditional black-box models. However, the possible loss of the predictive performance for gaining interpretability is often inevitable, putting practitioners in a dilemma of choosing between high accuracy (black-box models) and interpretability (interpr…
Paper proposes a new black-box adversarial attack using normalizing flows.
problem Adversarial vulnerability of deep neural networks.
method Proposes a novel black-box adversarial attack using normalizing flows.
result Demonstrates competitive performance against well-known black-box adversarial attack methods.
ODS improves adversarial attacks by maximizing output diversity.
problem Efficiency and effectiveness of adversarial attacks, especially black-box attacks.
method Output Diversified Sampling (ODS) that maximizes diversity in model outputs.
result ODS reduces the number of queries needed for black-box attacks on ImageNet by a factor of two.
Paper proposes ZO-NGD for more efficient black-box attacks.
problem Vulnerability of state-of-the-art DNNs to adversarial attacks.
method Zeroth-order natural gradient descent (ZO-NGD) for black-box attacks.
result ZO-NGD achieves significantly lower model query complexities.
Develops method to assess feature importance in black-box models for unconditional distribution.
problem Lack of methods to analyze feature importance in black-box models for unconditional distribution.
method Approximation method to compute feature importance curves for unconditional distribution.
result Produces sparse and faithful results, computationally efficient.
Paper introduces PRMs to learn non-Markovian stochastic rewards for reinforcement learning.
problem Lack of structured representation for non-Markovian stochastic rewards in reinforcement learning.
method Introduces probabilistic reward machines (PRMs) and presents an algorithm to learn them from decision processes.
result Algorithm proves correct and convergent for learning PRMs from decision processes.