Language models (LMs) trained on web-scale datasets are largely successful due to their ability to memorize large amounts of training data, even if only present in a few examples. These capabilities are often desirable in evaluation on tasks such as question answering but raise questions about whether these models can exhibit genuine reasoning or succeed only at mimicking patterns from the training data. This distinction is particularly salient in forecasting tasks, where the answer is not present in the training data, and the model must reason to make logical deductions. We present Reasoning and Tools for Forecasting (RTF), a framework of reasoning-and-acting (ReAct) agents that can dynamically retrieve updated information and run numerical simulation with equipped tools. We evaluate our model with questions from competitive forecasting platforms and demonstrate that our method is competitive with and can outperform human predictions. This suggests that LMs, with the right tools, can indeed think and adapt like humans, offering valuable insights for real-world decision-making.

Subword regularization, used widely in NLP, improves model performance by reducing the dependency on exact tokenizations, augmenting the training corpus, and exposing the model to more unique contexts during training. BPE and MaxMatch, two popular subword tokenization schemes, have stochastic dropout regularization variants. However, there has not been an analysis of the distributions formed by them. We show that these stochastic variants are heavily biased towards a small set of tokenizations per word. If the benefits of subword regularization are as mentioned, we hypothesize that biasedness artificially limits the effectiveness of these schemes. Thus, we propose an algorithm to uniformly sample tokenizations that we use as a drop-in replacement for the stochastic aspects of existing tokenizers, and find that it improves machine translation quality.

The Euclidean Shortest Path Problem (ESPP), which involves finding the shortest path in a Euclidean plane with polygonal obstacles, is a classic problem with numerous real-world applications. The current state-of-the-art solution, Euclidean Hub Labeling (EHL), offers ultra-fast query performance, outperforming existing techniques by 1-2 orders of magnitude in runtime efficiency. However, this performance comes at the cost of significant memory overhead, requiring up to tens of gigabytes of storage on large maps, which can limit its applicability in memory-constrained environments like mobile phones or smaller devices. Additionally, EHL's memory usage can only be determined after index construction, and while it provides a memory-runtime tradeoff, it does not fully optimize memory utilization. In this work, we introduce an improved version of EHL, called EHL*, which overcomes these limitations. A key contribution of EHL* is its ability to create an index that adheres to a specified memory budget while optimizing query runtime performance. Moreover, EHL* can leverage preknown query distributions, a common scenario in many real-world applications to further enhance runtime efficiency. Our results show that EHL* can reduce memory usage by up to 10-20 times without much impact on query runtime performance compared to EHL, making it a highly effective solution for optimal pathfinding in memory-constrained environments.

Multiple-choice question answering (MCQA) becomes particularly challenging when all choices are relevant to the question and are semantically similar. Yet this setting of MCQA can potentially provide valuable clues for choosing the right answer. Existing models often rank each choice separately, overlooking the context provided by other choices. Specifically, they fail to leverage the semantic commonalities and nuances among the choices for reasoning. In this paper, we propose a novel MCQA model by differentiating choices through identifying and eliminating their commonality, called DCQA. Our model captures token-level attention of each choice to the question, and separates tokens of the question attended to by all the choices (i.e., commonalities) from those by individual choices (i.e., nuances). Using the nuances as refined contexts for the choices, our model can effectively differentiate choices with subtle differences and provide justifications for choosing the correct answer. We conduct comprehensive experiments across five commonly used MCQA benchmarks, demonstrating that DCQA consistently outperforms baseline models. Furthermore, our case study illustrates the effectiveness of the approach in directing the attention of the model to more differentiating features.

Hierarchical methods represent state-of-the-art visual localization, optimizing search efficiency by using global descriptors to focus on relevant map regions. However, this state-of-the-art performance comes at the cost of substantial memory requirements, as all database images must be stored for feature matching. In contrast, direct 2D-3D matching algorithms require significantly less memory but suffer from lower accuracy due to the larger and more ambiguous search space. We address this ambiguity by fusing local and global descriptors using a weighted average operator within a 2D-3D search framework. This fusion rearranges the local descriptor space such that geographically nearby local descriptors are closer in the feature space according to the global descriptors. Therefore, the number of irrelevant competing descriptors decreases, specifically if they are geographically distant, thereby increasing the likelihood of correctly matching a query descriptor. We consistently improve the accuracy over local-only systems and achieve performance close to hierarchical methods while halving memory requirements. Extensive experiments using various state-of-the-art local and global descriptors across four different datasets demonstrate the effectiveness of our approach. For the first time, our approach enables direct matching algorithms to benefit from global descriptors while maintaining memory efficiency. The code for this paper will be published at \href{https://github.com/sontung/descriptor-disambiguation}{github.com/sontung/descriptor-disambiguation}.

This study addresses the challenge of fleet design optimization in the context of heterogeneous multi-robot fleets, aiming to obtain feasible designs that balance performance and costs. In the domain of autonomous multi-robot exploration, reinforcement learning agents play a central role, offering adaptability to complex terrains and facilitating collaboration among robots. However, modifying the fleet composition results in changes in the learned behavior, and training multi-robot systems using multi-agent reinforcement learning is expensive. Therefore, an exhaustive evaluation of each potential fleet design is infeasible. To tackle these hurdles, we introduce Bayesian Optimization for Fleet Design (BOFD), a framework leveraging multi-objective Bayesian Optimization to explore fleets on the Pareto front of performance and cost while accounting for uncertainty in the design space. Moreover, we establish a sub-linear bound for cumulative regret, supporting BOFD's robustness and efficacy. Extensive benchmark experiments in synthetic and simulated environments demonstrate the superiority of our framework over state-of-the-art methods, achieving efficient fleet designs with minimal fleet evaluations.

Breast ultrasound is essential for detecting and diagnosing abnormalities, with radiology reports summarizing key findings like lesion characteristics and malignancy assessments. Extracting this critical information is challenging due to the unstructured nature of these reports, with varied linguistic styles and inconsistent formatting. While proprietary LLMs like GPT-4 are effective, they are costly and raise privacy concerns when handling protected health information. This study presents a pipeline for developing an in-house LLM to extract clinical information from radiology reports. We first use GPT-4 to create a small labeled dataset, then fine-tune a Llama3-8B model on it. Evaluated on clinician-annotated reports, our model achieves an average F1 score of 84.6%, which is on par with GPT-4. Our findings demonstrate the feasibility of developing an in-house LLM that not only matches GPT-4's performance but also offers cost reductions and enhanced data privacy.

We present a probabilistic model for stochastic iterative algorithms with the use case of optimization algorithms in mind. Based on this model, we present PAC-Bayesian generalization bounds for functions that are defined on the trajectory of the learned algorithm, for example, the expected (non-asymptotic) convergence rate and the expected time to reach the stopping criterion. Thus, not only does this model allow for learning stochastic algorithms based on their empirical performance, it also yields results about their actual convergence rate and their actual convergence time. We stress that, since the model is valid in a more general setting than learning-to-optimize, it is of interest for other fields of application, too. Finally, we conduct five practically relevant experiments, showing the validity of our claims.

Global Climate Models (GCMs) are crucial for predicting future climate changes by simulating the Earth systems. However, GCM outputs exhibit systematic biases due to model uncertainties, parameterization simplifications, and inadequate representation of complex climate phenomena. Traditional bias correction methods, which rely on historical observation data and statistical techniques, often neglect unobserved confounders, leading to biased results. This paper proposes a novel bias correction approach to utilize both GCM and observational data to learn a factor model that captures multi-cause latent confounders. Inspired by recent advances in causality based time series deconfounding, our method first constructs a factor model to learn latent confounders from historical data and then applies them to enhance the bias correction process using advanced time series forecasting models. The experimental results demonstrate significant improvements in the accuracy of precipitation outputs. By addressing unobserved confounders, our approach offers a robust and theoretically grounded solution for climate model bias correction.

This paper introduces and assesses a cross-modal global visual localization system that can localize camera images within a color 3D map representation built using both visual and lidar sensing. We present three different state-of-the-art methods for creating the color 3D maps: point clouds, meshes, and neural radiance fields (NeRF). Our system constructs a database of synthetic RGB and depth image pairs from these representations. This database serves as the basis for global localization. We present an automatic approach that builds this database by synthesizing novel images of the scene and exploiting the 3D structure encoded in the different representations. Next, we present a global localization system that relies on the synthetic image database to accurately estimate the 6 DoF camera poses of monocular query images. Our localization approach relies on different learning-based global descriptors and feature detectors which enable robust image retrieval and matching despite the domain gap between (real) query camera images and the synthetic database images. We assess the system's performance through extensive real-world experiments in both indoor and outdoor settings, in order to evaluate the effectiveness of each map representation and the benefits against traditional structure-from-motion localization approaches. Our results show that all three map representations can achieve consistent localization success rates of 55% and higher across various environments. NeRF synthesized images show superior performance, localizing query images at an average success rate of 72%. Furthermore, we demonstrate that our synthesized database enables global localization even when the map creation data and the localization sequence are captured when travelling in opposite directions. Our system, operating in real-time on a mobile laptop equipped with a GPU, achieves a processing rate of 1Hz.