arXiv subjects

Victor M. Brea

Publications and source records attributed to Victor M. Brea.

2 recordsLinked to original sources

arXiv2026

Boosting the Memory Window of Memristive Stacks via Engineered Interfaces with High Ionic Mobility

Realizing the potential of oxide-based memristive devices for high-density data storage and energy-efficient computing still relies on overcoming key technical challenges, including the need for a larger number of stable resistance states, faster switching speeds, lower SET/RESET voltages, improved endurance, and reduced variability. Addressing these limitations requires innovative material design strategies. Here, we demonstrate that introducing a thin layer of oxide-ion conductor SrCoO3-x between the metal and the SrTiO3-based memristive elements expands the number of distinguishable resistance states from about 8 to about 22. This modification also reduced the SET/RESET voltage by 50% and markedly improved device endurance, albeit with a trade-off of reduced state retention. To assess the performance of this architecture, we trained a two-layer fully connected neural network using the experimental SrTiO3/SrCoO3-x memristor characteristics on the MNIST handwritten digit dataset. Networks with hidden-layer sizes between 64 and 256 neurons achieved classification errors below 7%. Finally, we confirmed the transferability of this interface-engineering approach by applying it to HfOx-based devices, achieving a consistent enhancement in the resistive state window.

cond-mat.mtrl-sci↗

arXiv2024

Lost and Found: Overcoming Detector Failures in Online Multi-Object Tracking

Multi-object tracking (MOT) endeavors to precisely estimate the positions and identities of multiple objects over time. The prevailing approach, tracking-by-detection (TbD), first detects objects and then links detections, resulting in a simple yet effective method. However, contemporary detectors may occasionally miss some objects in certain frames, causing trackers to cease tracking prematurely. To tackle this issue, we propose BUSCA, meaning `to search', a versatile framework compatible with any online TbD system, enhancing its ability to persistently track those objects missed by the detector, primarily due to occlusions. Remarkably, this is accomplished without modifying past tracking results or accessing future frames, i.e., in a fully online manner. BUSCA generates proposals based on neighboring tracks, motion, and learned tokens. Utilizing a decision Transformer that integrates multimodal visual and spatiotemporal information, it addresses the object-proposal association as a multi-choice question-answering task. BUSCA is trained independently of the underlying tracker, solely on synthetic data, without requiring fine-tuning. Through BUSCA, we showcase consistent performance enhancements across five different trackers and establish a new state-of-the-art baseline across three different benchmarks. Code available at: https://github.com/lorenzovaquero/BUSCA.

cs.CV↗