arXiv:2603.19899v1 Announce Type: new Abstract: Autocorrelation is a defining characteristic of time-series data, where each observation is statistically dependent on its predecessors. In the context of deep time-series forecasting, autocorrelation arises in both the input history and the label sequences, presenting two central research challenges: (1) designing neural architectures that model autocorrelation in history sequences, and (2) devising learning objectives that model autocorrelation in label sequences. Recent studies have made strides in tackling these challenges, but a systematic […]

Ver mais

I had about 940,000 PDFs to process. Running VLMs over a million pages is slow and expensive, and that gap is only getting worse as OCR moves toward transformer and VLM-based approaches. They’re great for complex understanding, but throughput and cost can become a bottleneck at scale. PaddleOCR (the non VL version), in my opinion the best non-VLM open source OCR, only handled ~15 img/s on my RTX 5090, which was still too slow. PaddleOCR-VL was crawling at […]

Ver mais

Selecting training parameters for convolutional neural networks (CNNs) and determining the amount of training data required for reliable generalization remain challenging and often time-consuming tasks, typically relying on manual trial-and-error. While genetic algorithms (GAs) have been applied to hyperparameter tuning, less attention has been given to how the proportion of training data influences generalization performance. In this study, we propose an interactive GA-based framework that simultaneously optimizes key training parameters and the image usage rate, defined as the […]

Ver mais

arXiv:2602.05051v1 Announce Type: new Abstract: Offline reinforcement learning (RL) aims to learn the optimal policy from a fixed dataset generated by behavior policies without additional environment interactions. One common challenge that arises in this setting is the out-of-distribution (OOD) error, which occurs when the policy leaves the training distribution. Prior methods penalize a statistical distance term to keep the policy close to the behavior policy, but this constrains policy improvement and may not completely prevent OOD actions. Another […]

Ver mais

arXiv:2603.26741v1 Announce Type: new Abstract: We study language-conditioned visual navigation (LCVN), in which an embodied agent is asked to follow a natural language instruction based only on an initial egocentric observation. Without access to goal images, the agent must rely on language to shape its perception and continuous control, making the grounding problem particularly challenging. We formulate this problem as open-loop trajectory prediction conditioned on linguistic instructions and introduce the LCVN Dataset, a benchmark of 39,016 trajectories and […]

Ver mais

arXiv:2601.22305v1 Announce Type: new Abstract: Automatic workflow generation is the process of automatically synthesizing sequences of LLM calls, tool invocations, and post-processing steps for complex end-to-end tasks. Most prior methods cast this task as an optimization problem with limited theoretical grounding. We propose to cast workflow generation as Bayesian inference over a posterior distribution on workflows, and introduce textbf{Bayesian Workflow Generation (BWG)}, a sampling framework that builds workflows step-by-step using parallel look-ahead rollouts for importance weighting and a […]

Ver mais

Even with the holidays coming up, the digital rights news doesn’t stop. Thankfully, EFF is here to keep you up-to-date with our EFFector newsletter! In our latest issue, we’re explaining why politicians latest attempts to ban VPNs is a terrible idea; asking supporters to file public comments opposing new rules that would make bad patents untouchable; and sharing a privacy victory—Sacramento is forced to end its dragnet surveillance program of power meter data. Prefer to listen in? Check out our audio companion, where […]

Ver mais