The performance of conventional interference management strategies degrades when interference power is comparable to signal power. We consider a new perspective on interference management using semantic communication. Specifically, a multi-user semantic communication system is considered on moderate interference channels (ICs), for which a novel framework of deep learning-based prompt-assisted semantic interference cancellation (DeepPASIC) is proposed. Each transmitted signal is partitioned into common and private parts. The common parts of different users are transmitted simultaneously in a shared medium, resulting in superposition. The private part, on the other hand, serves as a prompt to assist in canceling the interference suffered by the common part at the semantic level. Simulation results demonstrate that the proposed DeepPASIC outperforms conventional interference management strategies under moderate interference conditions.

Dominant pre-trained language models (PLMs) have been successful in high-quality natural language generation. However, the analysis of their generation is not mature: do they acquire generalizable linguistic abstractions, or do they simply memorize and recover substrings of the training data? Especially, few studies focus on domain-specific PLM. In this study, we pre-trained domain-specific GPT-2 models using a limited corpus of Japanese newspaper articles and quantified memorization of training data by comparing them with general Japanese GPT-2 models. Our experiments revealed that domain-specific PLMs sometimes "copy and paste" on a large scale. Furthermore, we replicated the empirical finding that memorization is related to duplication, model size, and prompt length, in Japanese the same as in previous English studies. Our evaluations are relieved from data contamination concerns by focusing on newspaper paywalls, which prevent their use as training data. We hope that our paper encourages a sound discussion such as the security and copyright of PLMs.

After Celebgate and the Fappening, a new iOS app called Nude is trying to protect iOS users' racy photos from another iCloud leak. The app scans a user's camera roll and pinpoints which photos, videos or documents include explicit content. After the AI scan, the images are locked into the app and erased to avoid hackers from getting their hands on the images. Jessica Chiu, Y.C. Chen and Edgar Khanzandian created the Nude app, which was released Oct. 4. "The app itself is very simple and intuitive to use," Chiu told International Business Times. "Once our proprietary AI technology scans through your camera roll and detects sensitive material, they are then imported into the app, deleted from your camera roll, and erased from iCloud."

Multi-agent planning using MA-STRIPS-related models is often motivated by the preservation of private information. Such motivation is not only natural for multi-agent systems, but it is one of the main reasons, why multi-agent planning (MAP) problems cannot be solved centrally. In this paper, we analyze privacy-preserving multi-agent planning (PP-MAP) from the perspective of secure multiparty computation (MPC). We discuss the concept of strong privacy and its implications and present two variants of a novel planner, provably strong privacy-preserving in general. As the main contribution, we formulate the limits of strong privacy-preserving planning in the terms of privacy, completeness and efficiency and show that, for a wide class of planning algorithms, all three properties are not achievable at once. Moreover, we provide a restricted variant of strong privacy based on equivalence classes of planning problems and show that an efficient, complete and strong privacy-preserving planner exists for such restriction.

"It shows that people respond to robots in a primitive, social way," researcher Jamy Li said in the release. "Social conventions regarding touching someone else's private parts apply to a robot's body parts as well. This research has implications for both robot design and theory of artificial systems." As shown in the video above, the robot in the experiment also provides an anatomical definition of each body part touched. There were four female volunteers and six male volunteers in the small study, according to The Guardian.

A study conducted by researchers at Stanford University and published in Science Daily found that humans get physiologically aroused by touching a robot's intimate areas. The researchers used Aldebaran Robotics' NAO, a humanoid robot, for their study. The robot instructed participants in the study to either touch or point at 13 different parts of its body. They then monitored the participants' responses as they carried out each command. The study found that participants more hesitant to touch more intimate parts of the robot, such as its eyes and buttocks, and they were physiologically aroused when touching these areas.

"It shows that people respond to robots in a primitive, social way," researcher Jamy Li said in the release. "Social conventions regarding touching someone else's private parts apply to a robot's body parts as well. This research has implications for both robot design and theory of artificial systems." As shown in the video above, the robot in the experiment also provides an anatomical definition of each body part touched. There were four female volunteers and six male volunteers in the small study, according to The Guardian.