Reinforcement Learning From Human Feedback (RLHF) has been a critical to the success of the latest generation of generative AI models. In response to the complex nature of the classical RLHF pipeline, direct alignment algorithms such as Direct Preference Optimization (DPO) have emerged as an alternative approach. Although DPO solves the same objective as the standard RLHF setup, there is a mismatch between the two approaches. Standard RLHF deploys reinforcement learning in a specific token-level MDP, while DPO is derived as a bandit problem in which the whole response of the model is treated as a single arm. In this work we rectify this difference, first we theoretically show that we can derive DPO in the token-level MDP as a general inverse Q-learning algorithm, which satisfies the Bellman equation. Using our theoretical results, we provide three concrete empirical insights. First, we show that because of its token level interpretation, DPO is able to perform some type of credit assignment. Next, we prove that under the token level formulation, classical search-based algorithms, such as MCTS, which have recently been applied to the language generation space, are equivalent to likelihood-based search on a DPO policy. Empirically we show that a simple beam search yields meaningful improvement over the base DPO policy. Finally, we show how the choice of reference policy causes implicit rewards to decline during training. We conclude by discussing applications of our work, including information elicitation in multi-tun dialogue, reasoning, agentic applications and end-to-end training of multi-model systems.

Rapidly Exploring Random Tree (RRT) algorithms, notably used for nonholonomic vehicle navigation in complex environments, are often not thoroughly evaluated for their specific challenges. This paper presents a first such comparison study of the variants Potential-Quick RRT* (PQ-RRT*), Informed RRT* (IRRT*), RRT*, and RRT, in maritime single-query nonholonomic motion planning. Additionally, the practicalities of using these algorithms in maritime environments are discussed and outlined. We also contend that these algorithms are beneficial not only for trajectory planning in Collision Avoidance Systems (CAS) but also for CAS verification when used as vessel behavior generators. Optimal RRT variants tend to produce more distance-optimal paths but require more computational time due to complex tree wiring and nearest neighbor searches. Our findings, supported by Welch`s t-test at a significance level of Alpha = 0.05, indicate that PQ-RRT* slightly outperform IRRT* and RRT* in achieving shorter trajectory length but at the expense of higher tuning complexity and longer run-times. Based on the results, we argue that these RRT algorithms are better suited for smaller-scale problems or environments with low obstacle congestion ratio. This is attributed to the curse of dimensionality, and trade-off with available memory and computational resources.

Pre-processing for whole slide images can affect classification performance both in the training and inference stages. Our study analyzes the impact of pre-processing parameters on inference and training across single- and multiple-domain datasets. However, searching for an optimal parameter set is time-consuming. To overcome this, we propose a novel Similarity-based Simulated Annealing approach for fast parameter tuning to enhance inference performance on single-domain data. Our method demonstrates significant performance improvements in accuracy, which raise accuracy from 0.512 to 0.847 in a single domain. We further extend our insight into training performance in multi-domain data by employing a novel Bayesian optimization to search optimal pre-processing parameters, resulting in a high AUC of 0.967. We highlight that better pre-processing for WSI can contribute to further accuracy improvement in the histology area.

We propose PyTorchGeoNodes, a differentiable module for reconstructing 3D objects from images using interpretable shape programs. In comparison to traditional CAD model retrieval methods, the use of shape programs for 3D reconstruction allows for reasoning about the semantic properties of reconstructed objects, editing, low memory footprint, etc. However, the utilization of shape programs for 3D scene understanding has been largely neglected in past works. As our main contribution, we enable gradient-based optimization by introducing a module that translates shape programs designed in Blender, for example, into efficient PyTorch code. We also provide a method that relies on PyTorchGeoNodes and is inspired by Monte Carlo Tree Search (MCTS) to jointly optimize discrete and continuous parameters of shape programs and reconstruct 3D objects for input scenes. In our experiments, we apply our algorithm to reconstruct 3D objects in the ScanNet dataset and evaluate our results against CAD model retrieval-based reconstructions. Our experiments indicate that our reconstructions match well the input scenes while enabling semantic reasoning about reconstructed objects.

We consider a variant of continuous-state partially-observable stochastic games with neural perception mechanisms and an asymmetric information structure. One agent has partial information, with the observation function implemented as a neural network, while the other agent is assumed to have full knowledge of the state. We present, for the first time, an efficient online method to compute an $\varepsilon$-minimax strategy profile, which requires only one linear program to be solved for each agent at every stage, instead of a complex estimation of opponent counterfactual values. For the partially-informed agent, we propose a continual resolving approach which uses lower bounds, pre-computed offline with heuristic search value iteration (HSVI), instead of opponent counterfactual values. This inherits the soundness of continual resolving at the cost of pre-computing the bound. For the fully-informed agent, we propose an inferred-belief strategy, where the agent maintains an inferred belief about the belief of the partially-informed agent based on (offline) upper bounds from HSVI, guaranteeing $\varepsilon$-distance to the value of the game at the initial belief known to both agents.

We introduce the Multi-Robot Connected Fermat Spiral (MCFS), a novel algorithmic framework for Multi-Robot Coverage Path Planning (MCPP) that adapts Connected Fermat Spiral (CFS) from the computer graphics community to multi-robot coordination for the first time. MCFS uniquely enables the orchestration of multiple robots to generate coverage paths that contour around arbitrarily shaped obstacles, a feature that is notably lacking in traditional methods. Our framework not only enhances area coverage and optimizes task performance, particularly in terms of makespan, for workspaces rich in irregular obstacles but also addresses the challenges of path continuity and curvature critical for non-holonomic robots by generating smooth paths without decomposing the workspace. MCFS solves MCPP by constructing a graph of isolines and transforming MCPP into a combinatorial optimization problem, aiming to minimize the makespan while covering all vertices. Our contributions include developing a unified CFS version for scalable and adaptable MCPP, extending it to MCPP with novel optimization techniques for cost reduction and path continuity and smoothness, and demonstrating through extensive experiments that MCFS outperforms existing MCPP methods in makespan, path curvature, coverage ratio, and overlapping ratio. Our research marks a significant step in MCPP, showcasing the fusion of computer graphics and automated planning principles to advance the capabilities of multi-robot systems in complex environments. Our code is available at https://github.com/reso1/MCFS.

For natural language understanding and generation, embedding concepts using an order-based representation is an essential task. Unlike traditional point vector based representation, an order-based representation imposes geometric constraints on the representation vectors for explicitly capturing various semantic relationships that may exist between a pair of concepts. In existing literature, several approaches on order-based embedding have been proposed, mostly focusing on capturing hierarchical relationships; examples include vectors in Euclidean space, complex, Hyperbolic, order, and Box Embedding. Box embedding creates region-based rich representation of concepts, but along the process it sacrifices simplicity, requiring a custom-made optimization scheme for learning the representation. Hyperbolic embedding improves embedding quality by exploiting the ever-expanding property of Hyperbolic space, but it also suffers from the same fate as box embedding as gradient descent like optimization is not simple in the Hyperbolic space. In this work, we propose Binder, a novel approach for order-based representation. Binder uses binary vectors for embedding, so the embedding vectors are compact with an order of magnitude smaller footprint than other methods. Binder uses a simple and efficient optimization scheme for learning representation vectors with a linear time complexity. Our comprehensive experimental results show that Binder is very accurate, yielding competitive results on the representation task. But Binder stands out from its competitors on the transitive closure link prediction task as it can learn concept embeddings just from the direct edges, whereas all existing order-based approaches rely on the indirect edges.

Vector data management systems (VDMSs) have become an indispensable cornerstone in large-scale information retrieval and machine learning systems like large language models. To enhance the efficiency and flexibility of similarity search, VDMS exposes many tunable index parameters and system parameters for users to specify. However, due to the inherent characteristics of VDMS, automatic performance tuning for VDMS faces several critical challenges, which cannot be well addressed by the existing auto-tuning methods. In this paper, we introduce VDTuner, a learning-based automatic performance tuning framework for VDMS, leveraging multi-objective Bayesian optimization. VDTuner overcomes the challenges associated with VDMS by efficiently exploring a complex multi-dimensional parameter space without requiring any prior knowledge. Moreover, it is able to achieve a good balance between search speed and recall rate, delivering an optimal configuration. Extensive evaluations demonstrate that VDTuner can markedly improve VDMS performance (14.12% in search speed and 186.38% in recall rate) compared with default setting, and is more efficient compared with state-of-the-art baselines (up to 3.57 times faster in terms of tuning time). In addition, VDTuner is scalable to specific user preference and cost-aware optimization objective. VDTuner is available online at https://github.com/tiannuo-yang/VDTuner.

Hardware-aware Neural Architecture Search approaches (HW-NAS) automate the design of deep learning architectures, tailored specifically to a given target hardware platform. Yet, these techniques demand substantial computational resources, primarily due to the expensive process of assessing the performance of identified architectures. To alleviate this problem, a recent direction in the literature has employed representation similarity metric for efficiently evaluating architecture performance. Nonetheless, since it is inherently a single objective method, it requires multiple runs to identify the optimal architecture set satisfying the diverse hardware cost constraints, thereby increasing the search cost. Furthermore, simply converting the single objective into a multi-objective approach results in an under-explored architectural search space. In this study, we propose a Multi-Objective method to address the HW-NAS problem, called MO-HDNAS, to identify the trade-off set of architectures in a single run with low computational cost. This is achieved by optimizing three objectives: maximizing the representation similarity metric, minimizing hardware cost, and maximizing the hardware cost diversity. The third objective, i.e. hardware cost diversity, is used to facilitate a better exploration of the architecture search space. Experimental results demonstrate the effectiveness of our proposed method in efficiently addressing the HW-NAS problem across six edge devices for the image classification task.

GPU kernels have come to the forefront of computing due to their utility in varied fields, from high-performance computing to machine learning. A typical GPU compute kernel is invoked millions, if not billions of times in a typical application, which makes their performance highly critical. Due to the unknown nature of the optimization surface, an exhaustive search is required to discover the global optimum, which is infeasible due to the possible exponential number of parameter combinations. In this work, we propose a methodology that uses deep sequence-to-sequence models to predict the optimal tuning parameters governing compute kernels. This work considers the prediction of kernel parameters as a sequence to the sequence translation problem, borrowing models from the Natural Language Processing (NLP) domain. Parameters describing the input, output and weight tensors are considered as the input language to the model that emits the corresponding kernel parameters. In essence, the model translates the problem parameter language to kernel parameter language. The core contributions of this work are: a) Proposing that a sequence to sequence model can accurately learn the performance dynamics of a GPU compute kernel b) A novel network architecture which predicts the kernel tuning parameters for GPU kernels, c) A constrained beam search which incorporates the physical limits of the GPU hardware as well as other expert knowledge reducing the search space. The proposed algorithm can achieve more than 90% accuracy on various convolutional kernels in MIOpen, the AMD machine learning primitives library. As a result, the proposed technique can reduce the development time and compute resources required to tune unseen input configurations, resulting in shorter development cycles, reduced development costs, and better user experience.