{"id":5158,"date":"2024-10-29T17:34:27","date_gmt":"2024-10-29T17:34:27","guid":{"rendered":"https:\/\/ai-grid.org\/?p=5158"},"modified":"2025-01-13T11:12:41","modified_gmt":"2025-01-13T11:12:41","slug":"ai-grid-auf-der-berlin-science-week-2024","status":"publish","type":"post","link":"https:\/\/ai-grid.org\/en\/event\/ai-grid-auf-der-berlin-science-week-2024\/","title":{"rendered":"AI Grid at Berlin Science Week 2024 \u2013 Experience the Next Generation of AI Research Up Close"},"content":{"rendered":"

On November 1st and 2nd, 2024, AI Grid will host a booth at the Museum f\u00fcr Naturkunde Berlin. Under this year's theme, 'Common Ground,' it invites die Berlin Science Week<\/a> visitors to explore new scientific approaches that can shape our world sustainably. For AI Grid, this means creating a space where young researchers present their innovative work in artificial intelligence (AI) and take visitors on a journey through the latest developments and applications in the field of AI.<\/p>\n\n\n\n

Meet the creative minds behind AI \u2013 and get inspired!<\/h3>\n\n\n\n

These three researchers demonstrate how AI is addressing current challenges in science and society:<\/p>\n\n\n\n

Mengisti Berihu Girmay<\/strong><\/h2>\n\n\n\n

Mengisti Berihu Girmay<\/strong> (Rhineland-Palatinate Technical University of Kaiserslautern-Landau) is working to make AI in agriculture more comprehensible and accessible. By focusing on transparent and explainable systems, she aims to build farmers' trust in AI technology and promote its practical application.<\/p>\n\n\n\n

What to Expect<\/strong><\/p>\n\n\n\n

In the field of AI-based technologies, decisions often appear opaque and difficult to understand. For farmers, who have relied on their own experience and intuition for years, this can understandably lead to skepticism. To address this issue, the concept of explainable AI (XAI) offers a promising solution.<\/strong><\/p>\n\n\n\n

However, before XAI can be implemented effectively, it is crucial to understand the specific needs of farmers regarding explainability. Which aspects of AI decisions require explanation? When and how should these explanations be presented to ensure they are clear and useful?<\/p>\n\n\n\n

By integrating these user-centered insights into AI systems, tools can be developed that make decision-making processes transparent and comprehensible. This approach aims not only to build trust but also to empower farmers to adopt AI technologies with confidence. Ultimately, this will simplify their daily work and drive innovation in the dairy industry.<\/p>\n\n\n\n

Hailay Kidu<\/strong><\/h2>\n\n\n\n

Hailay Kidu<\/strong> (L3S Research Center, Hannover) is researching technologies to improve access to digital information for Tigrinya speakers. His work breaks down language barriers and fosters cultural exchange by making Tigrinya more accessible in the digital space.<\/p>\n\n\n\n

What to Expect<\/strong><\/p>\n\n\n\n

We present a method to improve large language models (LLMs) in handling unknown words (out-of-vocabulary, OOV), tokenization challenges, and underrepresented languages. To achieve this, we expand the model's tokenization and embedding space by adding new tokens for these languages. These tokens are embedded in a way that aligns with the model's existing structure. This approach preserves the model's original performance while extending its support for additional languages.<\/p>\n\n\n\n

The method has been tested on languages using the Geez script but can also be applied to other underrepresented languages. The extended model demonstrates improved performance in tasks involving unknown words and enhanced tokenization. This approach provides a straightforward way to make LLMs more inclusive and functional for less widely spoken languages.<\/p>\n\n\n\n

Vivek Chavan<\/strong><\/h2>\n\n\n\n

Vivek Chavan<\/strong> (Fraunhofer IPK, Berlin) is developing energy-efficient machine learning models that continuously learn from new data without the need for constant retraining. His project, 'RECIL,' combines accuracy and energy efficiency, providing a sustainable solution for energy-intensive applications in industry.<\/p>\n\n\n\n

What to Expect<\/strong><\/p>\n\n\n\n

Imagine a pair of glasses that guides you through complex tasks, alerts you to mistakes, and provides real-time information. Vivek and his team are collaborating with Meta to develop intelligent wearable AI assistants like smart glasses. Their research focuses on egocentric vision\u2014teaching computers and AI models to perceive and understand the world from a human perspective. By collecting data and developing methods to capture environments, actions, and user intentions, they aim to support wearers in everyday and industrial scenarios. Smart glasses represent the next step in the evolution of smartphones and personal assistants, seamlessly integrating support into our daily lives and work environments.<\/p>\n\n\n\n

Boost your career \u2013 become part of AI Grid!<\/h3>\n\n\n\n

For master\u2019s students and PhD candidates specializing in AI, AI Grid provides the perfect platform to build a strong network early on, tailored to support your research and career. Connect directly with leading researchers and like-minded peers who share your interests. Through our Micro Focus Groups, you\u2019ll work on specific topics, receive personalized research guidance, and deepen your knowledge \u2013 all supported by experienced experts and professors.\n\nWith regular networking events across Germany and participation in our Science & Innovation Tours, you\u2019ll gain access to the European AI landscape and discover innovative approaches firsthand. AI Grid covers your travel expenses and offers talks by coaches and external partners to provide insights into current topics that foster your personal and professional growth.<\/p>\n\n\n\n

Join us to meet our researchers, experience AI Grid up close, and give your career a fresh boost. Apply now at www.ai-grid.org<\/a>to become part of this dynamic community!\"<\/p>\n\n\n\n

<\/span>