#Japan #AI Technology #Tokyo #NABLAS #JSAI 2026

NABLAS's Contributions at JSAI 2026

NABLAS Corporation, a leading AI research institute based in Tokyo, has announced its participation in the upcoming JSAI 2026 conference, set to take place in June 2026. This prestigious event is one of Japan's largest forums for artificial intelligence research. NABLAS will present three accepted papers, showcasing their latest advancements in the field of AI-generated videos.

Research Areas and Presentation Overview

The accepted papers revolve around critical themes such as fake video detection, a dedicated AI evaluation framework for Japanese culture, and a marketing support recommendation system based on food trends extracted from social media analysis. These research topics illustrate NABLAS's commitment to leveraging cutting-edge AI technologies to tackle societal challenges and foster new value creation.

Video Forgery Detection

With the rapid evolution of video generation technologies, detecting synthetic content has become increasingly complex. Traditional detection methods relied on identifying subtle inconsistencies between frames, but advancements in generative AI have made these discrepancies harder to spot. The research proposes a new detection framework that combines RGB appearance features with optical flow residuals to capitalize on spatial and temporal consistency in videos. Extensive experiments across various generative models demonstrated the robustness and effectiveness of this approach.

For further insights, visit Video Forgery Detection Paper.

Japanese Video-QA: A Cultural Emphasis

The Japanese Video-QA paper introduces a dedicated benchmark for video question answering focused on Japanese culture. While multi-modal large language models (MLLM) have made significant strides in processing images, audio, and video across languages, a robust evaluation framework specifically for Japanese cultural contexts has been lacking. This initiative aims to bridge that gap, highlighting the challenges associated with understanding video content influenced heavily by cultural nuances.

The dataset utilized consists of 800 question-answer pairs generated from 428 Japanese-related YouTube videos, covering various subjects such as seasonal events, tourist attractions, traditional culture, and pop culture.

To learn more, check out the complete details in the Japanese Video-QA Paper.

Marketing Support through Trend Prediction

Another key focus is the marketing support system, which uses social media analytics to identify emerging food trends. Sponsored by the Ministry of Economy, Trade and Industry and NEDO, this research dives deep into predicting trends through social media mentions. The system aims to provide insights into food products that might impact manufacturing, inventory, and logistics planning by identifying low-frequency and atypical food names.

This innovative approach seeks to enhance the extraction and organization of food trend candidates from social media discussions, ensuring businesses stay ahead of market changes.
Read more about this aspect in our Marketing Support Paper.

Collaboration with Toyota

Additionally, NABLAS will present two papers in collaboration with Toyota Motor Corporation, further solidifying their presence in the AI research landscape.

Upcoming Sessions

These presentations will take place at the JSAI 2026 conference on the following dates:

• - General Session: AI Applications in Marketing and Recommendations (June 8, 2026, 13:40 - 15:10, Venue: Conference Room 201B)
• - Poster Session: (June 8, 2026, 14:10 - 15:40, Venue: Exhibition Hall AB-1)

NABLAS's Vision and Mission

Founded in 2017 as a venture emerging from the University of Tokyo, NABLAS aims to provide cutting-edge AI educational content and solutions, particularly leveraging deep learning technologies. Their mission emphasizes the importance of AI in shaping a better future, and they continue to explore and innovate for societal benefit.

The company invites interested attendees at JSAI 2026 to discuss their research, with team members identifiable by their black NABLAS T-shirts. Feel free to approach and engage in insightful conversations about AI and its future applications.

For more information about NABLAS, visit their official website: NABLAS Corporation.