🎇✨🪔 Happy Diwali from NEC Laboratories America! 🪔✨🎇 This Sunday, colleagues, families, and friends worldwide celebrate the Festival of Lights. Diwali honors the triumph of light over darkness and knowledge over ignorance, a spirit that resonates with every researcher working to illuminate what comes next. Diwali dates back more than 2,500 years in India, where people lit oil lamps to welcome Lord Rama home to Ayodhya after his 14 years in exile. To everyone celebrating across our Princeton and San Jose teams and beyond: may this season bring you joy, prosperity, and bright new beginnings. #Diwali #HappyDiwali #FestivalOfLights

Applications are now open for Summer 2027 internships at NEC Labs America for students pursuing advanced degrees in Computer Science and Electrical Engineering, with positions available across all six research departments in Princeton, NJ, and San Jose, CA. Interns spend about three months working directly with senior researchers on cutting-edge projects in machine learning, computer vision, optical communications, and data science. Learn more: buff.ly/0r9cjbx #PHDLife #PHDJobs #STEM

Our news page features the latest research breakthroughs and announcements from across NEC Laboratories America and our six research departments. Recent stories include: ✔️ Meet Yue Tian: Leading Optical Sensing and Solutions at NEC Laboratories America (October 27): buff.ly/I0eWUCW ✔️ Routing Data Through Satellites That Never Stand Still (October 22): buff.ly/IhDLmx9 ✔️ Apply for a Summer 2027 Internship at NEC Labs America (October 21): buff.ly/iuEQJZ2 ✔️ Updating What AI Knows Without Breaking What It Already Knows (October 20): buff.ly/TdioW65 ✔️ Meet Yue-Kai Huang: Leading Our New Optical Communications & Computing Department (October 13): buff.ly/doHMuNK ✔️ Message from Chris White, New Head of NEC Research & Development (October 5): buff.ly/e1KHXGe ✔️ Christopher White Appointed Corporate Senior Vice President and Global Managing Director, R&D at NEC Corporation (September 23): buff.ly/hHNeh2T ✔️ Teaching AI to Edit Driving Scenes It Has Never Seen with HorizonWeaver (August 11): buff.ly/RKXXfd8 Visit our news page to stay current on what our research teams are working on: buff.ly/YiR7uYq #AppliedResearch #AI #Research

At NEC Labs America, research begins with hard problems, and our work spans six departments at the intersection of #ArtificialIntelligence, communication networks, and the physical world. We partner with colleagues across NEC, leading universities, and industry to advance research in data science, integrated systems, machine learning, media analytics, optical communications and computing, and optical sensing. Learn more: buff.ly/C3AzDaT #AppliedResearch #Innovation

CAViAR: A Causal Video Dataset for Fine-Grained Accident Reasoning in Real-World Scenarios Publication Date: 9/9/2026 Event: 6th DriveX Workshop in conjunction with #ECCV2026 Reference: pp. 1-16, 2026 Authors: Sparsh Garg, NEC Laboratories America, Inc.; Yi-Wen Chen, NEC Laboratories America, Inc.; Vijay Kumar B G, NEC Laboratories America, Inc.; Abhishek Aich, NEC Laboratories America, Inc. Abstract: While modern autonomous driving systems excel at perception tasks such as object detection and trajectory prediction, they lack the high-level causal reasoning required to interpret traffic accidents. In particular, determining responsibility, such as identifying who is at fault and which traffic rule was violated, remains largely unexplored in current benchmarks. To this end, we introduce CAViAR (Causal Accident Video and Incident Analysis Repository), a human-annotated dashcam benchmark comprising 2,249 real-world accident videos collected from CarCrashDataset (CCD) and Nexar. Each video is annotated with structured labels spanning environmental conditions, accident type, causal explanation, apparent At-Fault Agent, affected agent, and apparent rule-violation category. We benchmark state-of-the-art vision-language models (VLMs), including Cosmos-Reason2, Qwen3-VL, andInternVL3. Once class imbalance is accounted for with majority/random baselines and balanced metrics, perceptual competence is uneven–lighting. Learn more: nec-labs.com/blog/caviar-a-c… Tags: abhishek aich, #autonomousdriving, ma publication, media analytics, sparsh garg, vijay kumar b g, vision language models, yi wen chen

Our Optical Communications & Computing researchers publish influential work across optical transmission, intelligent networking, and photonic computing, from theory to industry-first field trials. Recent publications include: ✔️ Routing Data Through Satellites That Never Stand Still ✔️ Virtual Session ID-Based Handover for LEO Satellite Networks ✔️ Physical-Model-Based Optical Line System Transmission Optimization Integrating Probing Transceivers ✔️ Advancing Forward-Transmission Fibre Sensing for Precise Event Localisation Learn more: buff.ly/H0me1Zr #NECLabs #Research #OpticalCommunications #Photonics

Since 1988, NEC Labs America has served as the U.S.-based center for NEC Corporation’s global network of research laboratories, pursuing problem-driven research motivated by fundamental human needs across six specialized, multidisciplinary departments at our campuses in Princeton, NJ, and San Jose, CA. We seek hard problems at the forefront of knowledge, publish our results in top journals, and collaborate with industry, academia, and governments to turn disruptive solutions into real-world impact. Learn more: buff.ly/0ur44Tq #NECLabsAmerica #AppliedResearch #ResearchCommunity

🎉NEC Labs America turned 24 on November 1st! 🎉 This year, President Christopher White took on NEC’s global R&D leadership; we launched two new optical departments, and our researchers earned a 2026 Edison Patent Award. Add city-wide fiber sensing, smarter satellite routing, and more reliable AI, and it has been quite a year. Thank you to everyone who makes it possible! Read more about our successful year. nec-labs.com/blog/nec-labora… #NECLabs #Research #AI #OpticalNetworking

Our Machine Learning department has led #DeepLearning, support vector machines, and semantic analysis for over a decade, with current work spanning physics-informed machine learning, large multimodal healthcare models, and human-collaborative LLM agents. Our research aims to deliver more accurate medical diagnoses, safer autonomous systems, and more efficient energy use, with a steadfast commitment to explainability, fairness, and #EthicalAI. Learn more: buff.ly/4hsqe5Y #MachineLearning #NECLabsAmerica

🎃Happy Halloween from NEC Laboratories America! 🎃 Things that go byte in the night? Not on our watch. 🐛 Our Data Science & System Security team exorcises every bug lurking in the code. 🧠 Our Machine Learning researchers make sure neural networks never get ghosted by bad data. 🫕 Our Integrated Systems team brews a batch of blazing-fast breakthroughs, bubbling straight from the cauldron to the cloud. 💡 Our Optical Communications & Computing team computes with light, leaving no dark corners for latency to haunt. 🔦 Our Optical Sensing & Solutions team turns ordinary fiber into a sixth sense, detecting vibrations from miles away. 👻 Our Media Analytics team sees everything. Even the ghosts. Spooky? Maybe. Science? Definitely. No skeletons in our closet. Just skeleton code waiting for its next big discovery. Wishing everyone a boo-tiful night full of treats and zero crashes. ☠️☠️☠️ #HappyHalloween #NECLabsAmerica

Our Optical Sensing & Solutions department develops technologies that sense the physical world, transform observations into intelligence, and enable informed action, with research spanning distributed fiber sensing, infrared imaging, radio-frequency spectrum sensing, and AI-enabled sensing analytics. From early concepts and lab validation to proven field deployments on live infrastructure, our work helps create safer, more resilient, and more autonomous systems for infrastructure, utilities, transportation, environmental monitoring, and smart cities. Learn more: buff.ly/Sy8wtD1 #OpticalSensing #DistributedSensing #NECLabsAmerica

Our Integrated Systems department is building a new generation of #AI systems that reshape how organizations operate, innovate, and create value, with research centered on human-AI collaboration, business-intent understanding, continual adaptation, and governance. Current projects include agentic AI for enterprise work, grounded multimodal retrieval, and LLM-based code generation across the computing continuum, all designed for reliability at enterprise scale. Learn more: buff.ly/5uB3p5f #IntelligentSystems #NECLabsAmerica

Reward-Tilted On-Policy Distillation for Acoustic Grounding in Audio-Language Models Publication Date: 9/22/2026 Event: arXiv Reference: /abs/2609.28778v1 Authors: Kaiyang Li, NEC Laboratories America, Inc., University of Connecticut @UConn; @ShaoboHan1, NEC Laboratories America, Inc.; Yue Tian, NEC Laboratories America, Inc.; Shihao Ji, University of Connecticut Abstract: Audio-language models (ALMs) can exploit textual shortcuts to answer questions while overlooking acoustic evidence, weakening audio understanding. On-policy distillation (OPD) trains compact ALMs by supervising student-generated responses with teacher predictions, but does not explicitly distinguish acoustic support from linguistic predictability. We propose Reward-Tilted On-Policy Distillation (RT-OPD) to strengthen acoustic grounding. Given the same question and student-generated text, a frozen teacher predicts the next token with and without audio inputs. Their log-probability contrast defines a reward that reshapes the teacher distribution for reverse-KL distillation, emphasizing the additional evidence provided by audio. Across two compact students and three benchmarks, RT-OPD consistently outperforms Vanilla OPD. Experiments with silenced and replacement audio further suggest that RT-OPD strengthens the student’s reliance on acoustic evidence. Our 3B model achieves 72.72% accuracy on MMAU, the highest among the compared 3B models and competitive with several 7B and 8B models. Learn more: nec-labs.com/blog/reward-til… Tags: audio language models, #audioreasoning, contrastive teacher targets, kaiyang li, on policy distillation, optical sensing solutions, oss publication, shaobo han, shihao ji, university of connecticut, yue tian

From November 3 to 5, come discover the future of urban management with our platform 𝑪𝑪𝑶𝑪 at the @SmartCityexpo. We'll be waiting for you at 𝐒𝐭𝐚𝐧𝐝 𝐃𝟕𝟎 (𝐏𝐚𝐛𝐞𝐥𝐥𝐨́𝐧 𝟑), within the @LasRozasInnova space. Come visit us!

Del 3 al 5 de noviembre, ven a descubrir el futuro de la gestión urbana con nuestra plataforma 𝑪𝑪𝑶𝑪 en el @SmartCityexpo. Te esperamos en el 𝐒𝐭𝐚𝐧𝐝 𝐃𝟕𝟎 (𝐏𝐚𝐛𝐞𝐥𝐥𝐨́𝐧 𝟑), dentro del espacio de @LasRozasInnova. ¡Ven a visitarnos!

Media Analytics researchers publish influential work across computer vision, multimodal learning, and AI-driven media understanding — advancing how machines interpret visual and sensory data at scale. Recent publications: ✔️ HorizonForge: Driving Scene Editing with Any Trajectories and Any Vehicles ✔️ HorizonWeaver: Generalizable Multi-Level Semantic Editing for Driving Scenes ✔️ LangDriveCTRL: Natural Language Controllable Driving Scene Editing with Multi-Modal Agents ✔️ PhyCo: Learning Controllable Physical Priors for Generative Motion ✔️ Driving Video Retrieval for Complex Queries with Structured Grounding buff.ly/Pq9EdHT #NECLabs #Research #ComputerVision #MultimodalAI

How do you make #AIAgents trustworthy enough to deploy? Haifeng Chen, Head of Data Science and System Security at NEC Laboratories America, addresses that question in his keynote, “Trustworthy Enterprise Software Transformation,” at the Rocky Mountain Workshop on Trustworthy and Responsible AI, University of Colorado, Boulder, on November 4th. He will present our research on AI-powered software engineering assurance, a new paradigm that goes beyond code generation to recover, document, and preserve the underlying business intent of existing software systems. Learn more: nec-labs.com/blog/haifeng-ch… #TrustworthyAI #AgenticAI #ResponsibleAI

Our Optical Communications & Computing department advances high-capacity optical transmission and emerging photonic computing architectures, with research spanning coherent transmission techniques, AI-aided digital signal processing, software-defined optical networking, and photonic integrated chips. From theoretical modeling to world-first technology field trials, our work is helping shape the future of connectivity, data processing, and AI infrastructure across industries. Learn more: buff.ly/Pd5lLMX #OpticalNetworking #Photonics #NECLabsAmerica

Meet Dr. Yue Tian, Department Head of Optical Sensing and Solutions at NEC Laboratories America. His OSS team turns telecom fiber already in the ground into a sensing network for cities, power grids, and undersea infrastructure, using AI to interpret what the fiber detects. Read about his background, the department's research directions, and the industry trends behind them. Learn more: nec-labs.com/blog/meet-yue-t… #OpticalSensing #FiberSensing #SmartCities

Our San Jose team hit the racetrack! 🏁 We were thrilled to welcome colleagues from @NEC’s Japan Mobility business unit for an afternoon of laps, friendly rivalry and plenty of laughs at K1 Speed. Days like this remind us how much stronger our work becomes when we connect beyond the screen. Thank you to everyone who joined us on the grid! #NECLabs #TeamBuilding