Palo Alto Research Center

This paper explores the role of generative artificial intelligence (GAI) in addressing electrical engineering (EE) challenges. It introduces the evolution of Al technologies, from rule-based systems to large language models (LLMs), and provides practical insights into how engineers can utilize generative Al tools effectively. Key concepts such as transformers, prompt design, fine-tuning, and retrieval-augmented generation (RAG) are explained in accessible terms. Real-world examples demonstrate both the potential and limitations of Al in solving EE problems, emphasizing areas where human expertise remains essential. The paper advocates for domain-specific LLMs and enhanced Al integration with resources like IEEE Xplore® to improve output relevance and reliability. It concludes with a discussion on ethical considerations, hallucinations, digital twins, data security, and future directions including Al agents and multi-modal systems. The paper aims to empower engineers with foundational knowledge and guidance to responsibly leverage generative Al for innovation and productivity in the electrical engineering domain.

In the twilight of Moore’s law, optimizing program performance has emerged as a central focus in computer architecture research. Yet, high-level source optimization remains challenging due to the intricate nature of understanding code semantics. Our approach unifies machine learning techniques with established insights and tools from computer architecture to tackle the inherent challenges of high-level optimization. In this work, we introduce a framework that harnesses large language models (LLMs) for high-level program optimization. We curate a dataset of competitive C++ submissions, each accompanied by extensive unit tests to capture performance-improving patterns. To mitigate the variability of performance measurements, we develop an evaluation harness using the gem5 full-system simulator. Our results show a mean speedup of 6.86, outperforming the average human optimization of 3.66×. We also give an overview of subsequent work in this space, describing how LLM-driven optimization enables autonomously applying performance-improving edits across billions of lines of code in Google data centers.

Abstract The problem of reinforcement learning (RL) in an unknown nonlinear dynamical system is equivalent to the search for an optimal feedback law utilizing only data from the simulations/rollouts of the dynamical system. Most RL techniques search over a complex global nonlinear feedback parametrization making them suffer from high training times as well as variance. Instead, we advocate searching over a local feedback representation consisting of an open-loop sequence, and an associated optimal linear feedback law completely determined by the open-loop. We show that this alternate approach results in highly efficient training, the answers obtained are globally optimum, repeatable with negligible variance, and hence reliable, and the resulting closed performance is superior to global state-of-the-art RL techniques. Finally, if we replan, whenever required, which is feasible due to the fast and reliable local solution, it allows us to recover the optimal global feedback law.

The omnidirectional (360°) video format favors user interaction within an immersive virtual environment, in a manner that is closer to real-world situations. Several studies have been proposing computational models to predict salient regions for this type of media. Currently, strategies inspired by methods originally applied to 2D videos have been adapted to assess the performance of attention models for omnidirectional videos, ignoring the fact that immersive media provides the observer with a different experience when compared with conventional media. Furthermore, the lack of consolidated proposals for performance limits for 360° videos represents a relevant gap in the literature. This paper evaluates the applicability of 2D approaches (Mean Eye Position and Human Infinite), introduces Equator Bias as a performance bound, and also proposes Saliency Sum as an approach for estimating upper limits in visual attention models for immersive media. For validation purposes, we compare these bounds with outputs from three attention models (DAVE, Static CP360 and Spherical U-Net) for six metrics (AuC-Judd, AuC-Borji, NSS, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">KL</sub>, CC and SIM). The experiments were conducted on a subset of 6 videos from the PAVS10K dataset, which includes eye-tracking data from approximately 20 observers. In our experiments, the proposed Saliency Sum scored best on three metrics and Human Infinite achieved the best scores in two metrics and was never ranked as the worst. In contrast, Mean Eye Position showed the lowest performance, obtaining the worst score in five metrics and not ranking as the best in any. The Equator Bias showed worse score in one metric and scores close to the worst scores in three metrics. These results suggest that these approaches may serve as upper and lower performance bounds for evaluating saliency models in immersive videos.

Generative Artificial Intelligence (AI) models have risen to prominence due to their unparalleled ability to craft and generate digital assets, encompassing text, images, audio, video, and 3D models. Leveraging the capabilities of diffusion models, such as Stable Diffusion and Instruct pix2pix, users can guide AI with specific prompts, streamlining the creative journey for graphic designers. However, the primary application of these models has been to graphic content within desktop interfaces, prompting professionals in interior and architectural design to seek more tailored solutions for their daily operations. To bridge this gap, Augmented Reality (AR) and Mixed Reality (MR) technologies offer a promising solution, transforming traditional 2D artworks into engaging 3D interactive realms. In this paper, we present "Dream Mesh," a MR application MR tool that combines a Speech-to-3D generative workflow besed on DreamFusion model without relying on pre-existing 3D content libraries. This innovative system empowers users to express 3D content needs through natural language input, promising transformative potential in real-time 3D content creation and an enhanced MR user experience.

Cao et al. introduce PANDA, an AI model designed for the early detection of pancreatic ductal adenocarcinoma (PDAC) using non-contrast CT scans. While the model shows great promise, it faces several challenges. Notably, its training predominantly on East Asian datasets raises concerns about generalizability across diverse populations. Additionally, PANDA's ability to detect rare lesions, such as pancreatic neuroendocrine tumors (PNETs), could be improved by integrating other imaging modalities. High specificity is a strength, but it also poses risks of false positives, which may lead to unnecessary procedures and increased healthcare costs. Implementing a tiered diagnostic approach and expanding training data to include a wider demographic are essential steps for enhancing PANDA's clinical utility and ensuring its successful global implementation, ultimately shifting the focus from late diagnosis to proactive early detection.

We demonstrate for the first time a frequency-modulated continuous wave photonic integrated blue laser with sub-MHz linewidth. The laser exhibits frequency excursion greater than 900 MHz at up to 1 MHz triangular chirp rate with a nonlinearity of less than 1.5%.

An aqueous electrochemically controlled host-guest encapsulation system demonstrates a large and synthetically tunable redox entropy change. Electrochemical entropy is the basis for thermally regenerative electrochemical cycles (TRECs), which utilize reversible electrochemical processes with large molar entropy changes for thermogalvanic waste-heat harvesting and electrochemical cooling, among other potential applications. A supramolecular host-guest system demonstrates a molar entropy change of 4 times that of the state-of-the-art aqueous TREC electrolyte potassium ferricyanide. Upon encapsulation of a guest, water molecules that structurally resemble amorphous ice are displaced from the host cavity, leveraging a change in the degrees of freedom and ordering of the solvent rather than the solvation of the redox-active species to increase entropy. The synthetic tunability of the host allows rational optimization of the system's ΔS, showing a range of -51 to -101 cal mol-1 K-1 (-2.2 to -4.4 mV K-1) depending on ligand and metal vertex modifications, demonstrating the potential for rational design of high-entropy electrolytes and a new strategy to overcome theoretical limits on ion solvation reorganization entropy.

An experimental setup is designed to provide unsteady linear and rotational motion to a wing and measure aerodynamic loading. The focus of this investigation is a passive morphing wing that features a bi-stable trailing edge capable of snapping through instabilities to two different configurations in response to perceived pressure fields. The objectives of the study are three-fold: (i) to develop and characterize the experimental setup for future research in unsteady aerodynamics, (ii) to determine the aerodynamic loading experienced by the wing and capture the dynamic loading during the snap-through process, and (iii) to evaluate the performance of the morphing wing under varied conditions to enable improved iterations of the passive morphing design.