Foresee What You Will Learn: Data Augmentation for Domain Generalization in Non-stationary Environment
نویسندگان
چکیده
Existing domain generalization aims to learn a generalizable model perform well even on unseen domains. For many real-world machine learning applications, the data distribution often shifts gradually along indices. example, self-driving car with vision system drives from dawn dusk, sky darkening. Therefore, must be able adapt changes in ambient illuminations and continue drive safely road. In this paper, we formulate such problems as Evolving Domain Generalization, where generalize target by discovering leveraging evolving pattern of environment. We then propose Directional Augmentation (DDA), which simulates features mapping source augmentations through transformer. Specifically, DDA bi-level optimization problem solve it novel meta-learning approach representation space. evaluate proposed method both synthetic datasets datasets, empirical results show that our can outperform other existing methods.
منابع مشابه
Control Valve – What You Need to Learn
ii) The control valve Manufacturer can only state the Inherent valve characteristics, which refer to the flowrate (F) – valve travel position (m or Z) relationship, when the pressure drop across the control valve (DPV) is constant. Note that the pressure drop across the control valve (DPV) can be almost constant only if the piping system pressure drop (DPS) is concentrated at the control valve ...
متن کاملEvolution of Sociality: You Are What You Learn
Division of labour is the cornerstone of successful societies. A new study has shown that individual experience can produce long-lasting task specialisation in ants. This asks for a reappraisal of the role of individual learning in insect societies.
متن کاملWhat will I do without you?
In what case do you like reading so much? What about the type of the what will i do without you book? The needs to read? Well, everybody has their own reason why should read some books. Mostly, it will relate to their necessity to get knowledge from the book and want to read just to get entertainment. Novels, story book, and other entertaining books become so popular this day. Besides, the scie...
متن کاملPredicting What Will Happen When You Intervene
This paper offers some rules of thumb that practicing social workers can use for case studies that aim to construct, albeit not fully and never entirely reliably, models designed to help predict what will happen if they intervene in specific ways to help this particular client, here and now. We call these 'ex ante case-specific causal models'. 'Ex ante' because they are for before-the-fact pred...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2023
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v37i9.26320