Background
Developing machine/deep learning models that can demonstrate sophisticated reasoning is one of the most pressing challenges in AI research today. Basic mathematics has emerged as a key domain for evaluating progress in this area. The past few years have seen an explosion of neural network architectures, datasets, and benchmarks aimed at solving mathematical problems, achieving notable success in areas such as automated theorem proving, numerical integration, and the discovery of new algorithms. Yet, despite these advances, questions remain about whether deep learning models truly grasp fundamental concepts like quantities and symbolic numbers. This thesis will critically assess the current literature to determine whether state-of-the-art models can handle relatively simple arithmetic tasks. By examining their strengths and limitations, you will contribute to a deeper understanding of AI’s numerical proficiency.
What You Will Do
• Literature Review: Conduct a comprehensive review of existing research on how machine/deep learning models handle basic arithmetic tasks, identifying key strengths and gaps
• Benchmark Design: Develop tests and benchmarks to evaluate model performance in arithmetic, from basic calculations to complex reasoning
• Performance Analysis: Compare and analyze model accuracy and efficiency in arithmetic tasks, highlighting key insights
• Experimentation: Test various models using these benchmarks, analyzing their numerical reasoning capabilities
• Framework Development: Propose guidelines for assessing AI models’ numerical capabilities in future research
• Documentation and Presentation: Document your findings and present them to the MedixVision team and potentially in academic settings
What You Bring to the Table
• Currently enrolled as a full-time Bachelor’s student in Mathematics, Computer Science, or a related field, with a strong academic record (GPA: 3.00+/German Grade: 2,5)
• Proficient in academic writing, with a keen interest in research (publication of an academic article is an advantage)
• Experience in research methodologies, including conducting literature reviews and writing comprehensive reports
• Hands-on experience with machine/deep learning, particularly with neural networks
• Strong proficiency in English and/or German
What You Can Expect
• A supportive and collaborative working atmosphere within an international team.
• Professional supervision and support for your thesis, ensuring alignment with academic standards.
• Flexibility with home office options.
• Assistance with applications for Master’s programs at German universities, especially the Technical University of Munich.
Interested? Apply online now. We look forward to getting to know you! If you need information about this position, please contact us:
info[@]medixvision.com
•I have experience with an AI library in Python.
☐ Yes ☐ No
• I have experience finding specific information or data in the literature (e.g., ScienceDirect).
☐ Yes ☐ No
• I have a published academic paper, or I have contributed to a published study or work.
☐ Yes ☐ No
• An academic advisor from outside your university typically cannot supervise a thesis. Are there any professors at your university who can officially supervise this type of thesis?
☐ Yes ☐ No
• In which semester are you registered?
☐ 6 ☐ 7 ☐ 8 ☐ 8+
• I would like to pursue a Master’s degree abroad.
☐ Yes ☐ No
• I have EU citizenship or a residence permit.
☐ Yes ☐ No
What is your level of proficiency in English?
What is your level of proficiency in German?