Few-Shot Learning with Embedded Class Models and Shot-Free Meta Training (1905.04398v2)

Abstract: We propose a method for learning embeddings for few-shot learning that is suitable for use with any number of ways and any number of shots (shot-free). Rather than fixing the class prototypes to be the Euclidean average of sample embeddings, we allow them to live in a higher-dimensional space (embedded class models) and learn the prototypes along with the model parameters. The class representation function is defined implicitly, which allows us to deal with a variable number of shots per each class with a simple constant-size architecture. The class embedding encompasses metric learning, that facilitates adding new classes without crowding the class representation space. Despite being general and not tuned to the benchmark, our approach achieves state-of-the-art performance on the standard few-shot benchmark datasets.

Overview of the ICCV Author Guidelines in LaTeX

The paper "LaTeX Author Guidelines for ICCV Proceedings" delineates comprehensive instructions for authors preparing manuscripts for the International Conference on Computer Vision (ICCV). These guidelines are critical for ensuring consistent formatting and presentation across submissions, facilitating the peer review process, and maintaining the publication quality of ICCV proceedings.

Key Aspects of the Guidelines

The document places significant emphasis on several components of manuscript preparation:

• Language and Formatting: The paper mandates that all submissions be in English and adhere to specific formatting requirements, including the use of Times or Times Roman fonts, two-column layout, and defined margin specifications. The formatting ensures clarity and professional presentation.
• Paper Length and Review Process: Authors must adhere to an eight-page limit, excluding references, to facilitate efficient reviewing. Overlength papers will be rejected without review. This stringent rule underscores the importance of concise scientific communication.
• Blind Review Anonymity: To preserve the impartiality of the review process, authors are instructed on maintaining anonymity by avoiding personal pronouns in citations and referencing their prior work professionally. This requirement is crucial for maintaining the integrity of double-blind reviewing.
• Section and Equation Numbering: For ease of reference, all sections and displayed equations must be numbered. This aids readers and reviewers in pinpointing specific content within the manuscript, enhancing the navigation through complex scientific discussions.

Practical Implications

The guidelines serve both practical and theoretical functions. Practically, they provide a standardized template that simplifies the submission process for authors and ensures uniform presentation throughout ICCV publications. Theoretically, adherence to such stringent formatting promotes discipline in scientific writing, encouraging authors to present their ideas succinctly and clearly.

Potential Future Developments

As the field of AI and computer vision evolves, future refinements of these guidelines may encompass accommodating new technologies in manuscript preparation tools, potentially integrating automated compliance checks to streamline the submission process further. Enhancements in graphical representation guidelines could be anticipated, addressing challenges in visualizing complex data accurately and effectively within the constraints of traditional paper formats.

In conclusion, the author guidelines for ICCV proceedings are foundational in maintaining the high standards expected in AI conference publications. They ensure each submission is presented consistently, allowing for fair evaluation and facilitating scientific discourse within the computer vision community.