Active Human Feedback Collection via Neural Contextual Dueling Bandits

Collecting human preference feedback is often expensive, leading recent works to develop principled algorithms to select them more efficiently. However, these works assume that the underlying reward function is linear, an assumption that does not hold in many real-life applications, such as online recommendation and LLM alignment. To address this limitation, we propose Neural-ADB, an algorithm based on the neural contextual dueling bandit framework that provides a principled and practical method for collecting human preference feedback when the underlying latent reward function is non-linear. We theoretically show that when preference feedback follows the Bradley-Terry-Luce model, the worst sub-optimality gap of the policy learned by Neural-ADB decreases at a sub-linear rate as the preference dataset increases. Our experimental results on problem instances derived from synthetic preference datasets further validate the effectiveness of Neural-ADB.

Superfast Configuration-Space Convex Set Computation on GPUs for Online Motion Planning

In this work, we leverage GPUs to construct probabilistically collision-free convex sets in robot configuration space on the fly. This extends the use of modern motion planning algorithms that leverage such representations to changing environments. These planners rapidly and reliably optimize high-quality trajectories, without the burden of challenging nonconvex collision-avoidance constraints. We present an algorithm that inflates collision-free piecewise linear paths into sequences of convex sets (SCS) that are probabilistically collision-free using massive parallelism. We then integrate this algorithm into a motion planning pipeline, which leverages dynamic roadmaps to rapidly find one or multiple collision-free paths, and inflates them. We then optimize the trajectory through the probabilistically collision-free sets, simultaneously using the candidate trajectory to detect and remove collisions from the sets. We demonstrate the efficacy of our approach on a simulation benchmark and a KUKA iiwa 7 robot manipulator with perception in the loop. On our benchmark, our approach runs 17.1 times faster and yields a 27.9% increase in reliability over the nonlinear trajectory optimization baseline, while still producing high-quality motion plans.

Low Stein Discrepancy via Message-Passing Monte Carlo

Message-Passing Monte Carlo (MPMC) was recently introduced as a novel low-discrepancy sampling approach leveraging tools from geometric deep learning. While originally designed for generating uniform point sets, we extend this framework to sample from general multivariate probability distributions with known probability density function. Our proposed method, Stein-Message-Passing Monte Carlo (Stein-MPMC), minimizes a kernelized Stein discrepancy, ensuring improved sample quality. Finally, we show that Stein-MPMC outperforms competing methods, such as Stein Variational Gradient Descent and (greedy) Stein Points, by achieving a lower Stein discrepancy.

Enhancing Steering Estimation with Semantic-Aware GNNs

Steering estimation is a critical task in autonomous driving, traditionally relying on 2D image-based models. In this work, we explore the advantages of incorporating 3D spatial information through hybrid architectures that combine 3D neural network models with recurrent neural networks (RNNs) for temporal modeling, using LiDAR-based point clouds as input. We systematically evaluate four hybrid 3D models, all of which outperform the 2D-only baseline, with the Graph Neural Network (GNN) - RNN model yielding the best results. To reduce reliance on LiDAR, we leverage a pretrained unified model to estimate depth from monocular images, reconstructing pseudo-3D point clouds. We then adapt the GNN-RNN model, originally designed for LiDAR-based point clouds, to work with these pseudo-3D representations, achieving comparable or even superior performance compared to the LiDAR-based model. Additionally, the unified model provides semantic labels for each point, enabling a more structured scene representation. To further optimize graph construction, we introduce an efficient connectivity strategy where connections are predominantly formed between points of the same semantic class, with only 20\% of inter-class connections retained. This targeted approach reduces graph complexity and computational cost while preserving critical spatial relationships. Finally, we validate our approach on the KITTI dataset, achieving a 71% improvement over 2D-only models. Our findings highlight the advantages of 3D spatial information and efficient graph construction for steering estimation, while maintaining the cost-effectiveness of monocular images and avoiding the expense of LiDAR-based systems.

TETRIS: Optimal Draft Token Selection for Batch Speculative Decoding

We propose TETRIS, a novel method that optimizes the total throughput of batch speculative decoding in multi-request settings. Unlike existing methods that optimize for a single request or a group of requests as a whole, TETRIS actively selects the most promising draft tokens (for every request in a batch) to be accepted when verified in parallel, resulting in fewer rejected tokens and hence less wasted computing resources. Such an effective resource utilization to achieve fast inference in large language models (LLMs) is especially important to service providers with limited inference capacity. Compared to baseline speculative decoding, TETRIS yields a consistently higher acceptance rate and more effective utilization of the limited inference capacity. We show theoretically and empirically that TETRIS outperforms baseline speculative decoding and existing methods that dynamically select draft tokens, leading to a more efficient batch inference in LLMs.

Data value estimation on private gradients

For gradient-based machine learning (ML) methods commonly adopted in practice such as stochastic gradient descent, the de facto differential privacy (DP) technique is perturbing the gradients with random Gaussian noise. Data valuation attributes the ML performance to the training data and is widely used in privacy-aware applications that require enforcing DP such as data pricing, collaborative ML, and federated learning (FL). Can existing data valuation methods still be used when DP is enforced via gradient perturbations? We show that the answer is no with the default approach of injecting i.i.d.~random noise to the gradients because the estimation uncertainty of the data value estimation paradoxically linearly scales with more estimation budget, producing estimates almost like random guesses. To address this issue, we propose to instead inject carefully correlated noise to provably remove the linear scaling of estimation uncertainty w.r.t.~the budget. We also empirically demonstrate that our method gives better data value estimates on various ML tasks and is applicable to use cases including dataset valuation and~FL.

The Master Key Filters Hypothesis: Deep Filters Are General in DS-CNNs

This paper challenges the prevailing view that convolutional neural network (CNN) filters become increasingly specialized in deeper layers. Motivated by recent observations of clusterable repeating patterns in depthwise separable CNNs (DS-CNNs) trained on ImageNet, we extend this investigation across various domains and datasets. Our analysis of DS-CNNs reveals that deep filters maintain generality, contradicting the expected transition to class-specific filters. We demonstrate the generalizability of these filters through transfer learning experiments, showing that frozen filters from models trained on different datasets perform well and can be further improved when sourced from larger datasets. Our findings indicate that spatial features learned by depthwise separable convolutions remain generic across all layers, domains, and architectures. This research provides new insights into the nature of generalization in neural networks, particularly in DS-CNNs, and has significant implications for transfer learning and model design.

Generating Out-Of-Distribution Scenarios Using Language Models

The deployment of autonomous vehicles controlled by machine learning techniques requires extensive testing in diverse real-world environments, robust handling of edge cases and out-of-distribution scenarios, and comprehensive safety validation to ensure that these systems can navigate safely and effectively under unpredictable conditions. Addressing Out-Of-Distribution (OOD) driving scenarios is essential for enhancing safety, as OOD scenarios help validate the reliability of the models within the vehicle's autonomy stack. However, generating OOD scenarios is challenging due to their long-tailed distribution and rarity in urban driving dataset. Recently, Large Language Models (LLMs) have shown promise in autonomous driving, particularly for their zero-shot generalization and common-sense reasoning capabilities. In this paper, we leverage these LLM strengths to introduce a framework for generating diverse OOD driving scenarios. Our approach uses LLMs to construct a branching tree, where each branch represents a unique OOD scenario. These scenarios are then simulated in the CARLA simulator using an automated framework that aligns scene augmentation with the corresponding textual descriptions. We evaluate our framework through extensive simulations, and assess its performance via a diversity metric that measures the richness of the scenarios. Additionally, we introduce a new "OOD-ness" metric, which quantifies how much the generated scenarios deviate from typical urban driving conditions. Furthermore, we explore the capacity of modern Vision-Language Models (VLMs) to interpret and safely navigate through the simulated OOD scenarios. Our findings offer valuable insights into the reliability of language models in addressing OOD scenarios within the context of urban driving.

Learning autonomous driving from aerial imagery

In this work, we consider the problem of learning end to end perception to control for ground vehicles solely from aerial imagery. Photogrammetric simulators allow the synthesis of novel views through the transformation of pre-generated assets into novel views.However, they have a large setup cost, require careful collection of data and often human effort to create usable simulators. We use a Neural Radiance Field (NeRF) as an intermediate representation to synthesize novel views from the point of view of a ground vehicle. These novel viewpoints can then be used for several downstream autonomous navigation applications. In this work, we demonstrate the utility of novel view synthesis though the application of training a policy for end to end learning from images and depth data. In a traditional real to sim to real framework, the collected data would be transformed into a visual simulator which could then be used to generate novel views. In contrast, using a NeRF allows a compact representation and the ability to optimize over the parameters of the visual simulator as more data is gathered in the environment. We demonstrate the efficacy of our method in a custom built mini-city environment through the deployment of imitation policies on robotic cars. We additionally consider the task of place localization and demonstrate that our method is able to relocalize the car in the real world.

Flex: End-to-End Text-Instructed Visual Navigation with Foundation Models

End-to-end learning directly maps sensory inputs to actions, creating highly integrated and efficient policies for complex robotics tasks. However, such models are tricky to efficiently train and often struggle to generalize beyond their training scenarios, limiting adaptability to new environments, tasks, and concepts. In this work, we investigate the minimal data requirements and architectural adaptations necessary to achieve robust closed-loop performance with vision-based control policies under unseen text instructions and visual distribution shifts. To this end, we design datasets with various levels of data representation richness, refine feature extraction protocols by leveraging multi-modal foundation model encoders, and assess the suitability of different policy network heads. Our findings are synthesized in Flex (Fly-lexically), a framework that uses pre-trained Vision Language Models (VLMs) as frozen patch-wise feature extractors, generating spatially aware embeddings that integrate semantic and visual information. These rich features form the basis for training highly robust downstream policies capable of generalizing across platforms, environments, and text-specified tasks. We demonstrate the effectiveness of this approach on quadrotor fly-to-target tasks, where agents trained via behavior cloning on a small simulated dataset successfully generalize to real-world scenes, handling diverse novel goals and command formulations.