Project

This research looks at how to make computer vision machine learning models more efficient. Inspired by green and sustainable practices, we’re not just trying to limit how we train these models. Instead, we’re looking at how to use our existing resources and information better. Our goal is to start a new approach called zero-waste machine learning that aims to save on computational efforts and use fewer resources. It relies on three pillars: sustainable machine learning for autonomous machines, knowledge accumulation in continually trained models and conditional computations.

General must-have requirements

If you are a student or graduate of Mathematics, Computer Science, Information Technology or a related discipline and would like to pursue a career in research, then share your plans with us by applying for this position.

Team leader - Conditional Computations

As a team leader you will be responsible to oversee the development of a team consisting of PhD students and post-docs that will work in the area of conditional computations (a.k.a. dynamic computations).

Must-have requiremets

Educational background: PhD in Science, preferred majors: Computer Science, or Mathematics or other science majors with experience in working with AI, experience in machine learning, English language skills at an advanced level, very good programming skills, a proactive approach to solving scientific problems and issues, strong analytical and critical thinking skills, ability to work in a team.

Nice-to-have requiremets

Scientific track record of publications presented during the most important thematic conferences such as: NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, CoRL, ACL, EMNLP, KDD, AAAI, IJCAI, COLT, UAI, AISTATS or in scientific journals TPAMI, IJCV, etc. Experience as a team leader is a plus.

PostDoc

As a post-doc, you will carry out research related to the following topics: efficient methods of machine learning in applications related to computer vision, algorithms of creating efficient representation of data, including multimodal data, as well as methods of machine learning concentrated on long-term development of artificial intelligence (continual learning, methods with reduced supervision including self-supervision methods, non-supervised methods and generative models).

Must-have requiremets

Scientific track record of publications presented during the most important thematic conferences such as: NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, CoRL, ACL, EMNLP, KDD, AAAI, IJCAI, COLT, UAI, AISTATS or in scientific journals TPAMI, IJCV, etc.

Nice-to-have requiremets

Scientific track record of publications presented during the most important thematic conferences such as: NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, CoRL, ACL, EMNLP, KDD, AAAI, IJCAI, COLT, UAI, AISTATS or in scientific journals TPAMI, IJCV, etc.,

PhD Student

As a PhD student, you will carry out state-of-the-art research on topics related to: performance-based machine learning, deep neural networks, continual learning, conditional computations, sustainable machine learning.

Must-have requiremets

A degree (Master’s or equivalent) in computer science, information technology or a related discipline from a leading university, excellent programming skills, curiosity and great interest in machine learning and/or artificial intelligence, fluency in English.

Nice-to-have requiremets

Scientific track record of publications presented during the most important thematic conferences such as: NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, CoRL, ACL, EMNLP, KDD, AAAI, IJCAI, COLT, UAI, AISTATS or in scientific journals TPAMI, IJCV, etc.,

Contact: Tomasz Trzciński (tomasz.trzcinski [ at ] ideas-ncbr.pl)

Project's lab: