Khoi P. N. Nguyen, Terrence Li, Derek Lou Zhou, Gabriel Xiong, Pranav Balu, Nandhan Alahari, Alan Huang, Tanush Chauhan, Harshavardhan Bala, Emre Guzelordu, Affan Kashfi, Aaron Xu, Suyesh Shrestha, Megan Vu, Jerry Wang, Vincent Ng. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Deepfake technology has advanced rapidly in recent years, enabling the creation of hyper-realistic manipulated media that can be weaponized for misinformation, fraud, and identity abuse. While generative models for face swapping and reenactment have improved, deepfake detection models often struggle in real-world deployments due to a multitude of problems. Some of them are due to constant advancements in attacks, domain shifts from generation, and more. This work explores Synthetic Domain Alignment (SDA) as a strategy for improving deepfake detection robustness. SDA leverages an unconditional generative model to synthesize new samples in both the source and target domains, then uses these synthetic datasets to retrain the detector with better domain alignment. This synthetic-to-synthetic alignment process helps bridge the gap between domains without requiring direct access to the original source data, making it suitable for privacy-sensitive or streaming applications. We evaluate our approach against a strong static baseline trained on FaceForensics++ under both in-domain and cross-dataset settings, including compressed, degraded, and unseen manipulation types. Experimental results demonstrate that SDA can significantly improve detection accuracy under severe domain shifts, highlighting its potential for practical deployment.

Zhiwei Fei, Xiaoyu Shen, Dawei Zhu, Fengzhe Zhou, Zhuo Han, Alan Huang, Songyang Zhang, Kai Chen, Zhixin Yin, Zongwen Shen, Jidong Ge, Vincent Ng. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Pointing is a typical means of directing a human's attention to a specific object or event. Robot pointing behaviours that direct the attention of humans are critical for human-robot interaction, communication and collaboration. In this paper, we describe an experiment undertaken to investigate human comprehension of a humanoid robot's pointing behaviour. We programmed a NAO robot to point to markers on a large screen and asked untrained human subjects to identify the target of the robots pointing gesture. We found that humans are able to identify robot pointing gestures. Human subjects achieved higher levels of comprehension when the robot pointed at objects closer to the gesturing arm and when they stood behind the robot. In addition, we found that subjects performance improved with each assessment task. These new results can be used to guide the design of effective robot pointing behaviours that enable more effective robot to human communication and improve human-robot collaborative performance.

Article Free Access Share on Optical digital computers Author: A. Huang AT&T Bell Laboratories, Crawfords Corner Road, 4G-514, Holmdel, New Jersey AT&T Bell Laboratories, Crawfords Corner Road, 4G-514, Holmdel, New JerseyView Profile Authors Info & Claims Supercomputing '89: Proceedings of the 1989 ACM/IEEE conference on SupercomputingAugust 1989 Pages 446–449https://doi.org/10.1145/76263.76312Published:01 August 1989Publication History 2citation922DownloadsMetricsTotal Citations2Total Downloads922Last 12 Months6Last 6 weeks2 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my Alerts New Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

The authors discuss digital optics, a technology for processing, transport, and storage of optical digital information. Digital optics offers both the high temporal bandwidth of fiber communications and the high connectivity and information density of optical imaging. The energy dissipation per bit of communicated information, as well as the chip area dedicated to interconnections, can be significantly lower in optics than in high-speed electronics. This motivates the introduction of parallel optical interconnections through free space in communication-intensive areas of digital information processing such as switching in telecommunications and within multiprocessors. Digital optical circuits can be constructed by cascading two-dimensional planar arrays of optical logic gates interconnected in free space. The state of the art and the trends in digital optical information processing systems for optical logic, optoelectronic interfaces, and optical free-space interconnection systems are reviewed.>

The residue number system is used to partition addition, subtraction, multiplication, or integer polynomial transforms into several simpler calculations, each of which can be processed in parallel with complete independence. These segments are computationally simple such that all the arithmetic interactions can be enumerated as mathematical mappings. By routing signals through various mappings a number may be encoded into residue form, processed in various ways, and eventually decoded back to a normal number system. The signals may be routed in a manner to reflect calculations involving a plurality of operands and operations. By routing several signals in close sequence calculations may be pipelined. By routing different types of signals independently several calculations may be multiplexed. Detection of abnormalities in a segment's signal may be used to exclude the segment from the decoding process, thus preserving the correctness of the calculations. By reducing and conveniently deferring the signal detections necessary, a greater variety of phenomena may be considered for digital computation. Optical implementations are of particular interest.