Did Full Self-Driving (Supervised), Tesla's driver assistance feature, play a role in a woman's death? On a Texas evening last week, a 76-year-old grandmother named Martha Avila was standing in the front room of her suburban home when a Tesla Model 3 hurtled into her brick home at a reported speed of over 70 miles per hour, killing her. The car's driver, 44-year-old Michael Butler, later told police that he had Tesla's driver assistance features --which the automaker argues make driving safer and less stressful--engaged during the crash. Butler exhibited "no signs of intoxication," the Harris County Sheriff's Office, which responded to the crash, noted in a report. Now Avila's family is suing not only Butler but also Tesla, alleging that the electric-auto maker's Full Self-Driving (Supervised) driver assistance feature, also called FSD, played a role in her death.

Photorealistic Codec Avatars (PCA), which generate high-fidelity human face renderings, are increasingly being used in Virtual Reality (VR) environments to enable immersive communication and interaction through deep learning-based generative models. However, these models impose significant computational demands, making real-time inference challenging on resource-constrained VR devices such as head-mounted displays (HMDs), where latency and power efficiency are critical. To address this challenge, we propose an efficient post-training quantization (PTQ) method tailored for Codec Avatar models, enabling low-precision execution without compromising output quality. In addition, we design a custom hardware accelerator that can be integrated into the system-on-chip (SoC) of VR devices to further enhance processing efficiency. Building on these components, we introduce ESCA, a full-stack optimization framework that accelerates PCA inference on edge VR platforms. Experimental results demonstrate that ESCA boosts FovVideoVDP quality scores by up to +0.39 over the best 4-bit baseline, delivers up to 3.36 latency reduction, and sustains a rendering rate of 100 frames per second in endto-end tests, satisfying real-time VR requirements. These results demonstrate the feasibility of deploying high-fidelity codec avatars on resource-constrained devices, opening the door to more immersive and portable VR experiences. Paper website can be found at https://zmzfpc.github.io/ESCA/.

Power efficiency is plateauing in the standard digital electronics realm such that new hardware, models, and algorithms are needed to reduce the costs of AI training. The combination of energy-based analog circuits and the Equilibrium Propagation (EP) algorithm constitutes a compelling alternative compute paradigm for gradient-based optimization of neural nets. Existing analog hardware accelerators, however, typically incorporate digital circuitry to sustain auxiliary non-weight-stationary operations, mitigate analog device imperfections, and leverage existing digital platforms. Such heterogeneous hardware lacks a supporting theoretical framework. In this work, we introduce \emph{Feedforward-tied Energy-based Models} (ff-EBMs), a hybrid model comprised of feedforward and energy-based blocks housed on digital and analog circuits. We derive a novel algorithm to compute gradients end-to-end in ff-EBMs by backpropagating and ``eq-propagating'' through feedforward and energy-based parts respectively, enabling EP to be applied flexibly on realistic architectures. We experimentally demonstrate the effectiveness of this approach on ff-EBMs using Deep Hopfield Networks (DHNs) as energy-based blocks, and show that a standard DHN can be arbitrarily split into any uniform size while maintaining or improving performance with increases in simulation speed of up to four times. We then train ff-EBMs on ImageNet32 where we establish a new state-of-the-art performance for the EP literature (46 top-1 \%). Our approach offers a principled, scalable, and incremental roadmap for the gradual integration of self-trainable analog computational primitives into existing digital accelerators.

Neural Information Processing Systems http://nips.cc/

Tiny machine learning (TinyML) aims to run ML models on small devices and is increasingly favored for its enhanced privacy, reduced latency, and low cost.

While inference on analog accelerators has been studied recently, the training perspective is underexplored. Recent studies have shown that the "workhorse" of digital AI training - stochastic gradient descent (

Neural Information Processing Systems http://nips.cc/