We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Understanding Dynamics of Nonlinear Representation Learning and Its Application

Abstract: Representations of the world environment play a crucial role in artificial intelligence. It is often inefficient to conduct reasoning and inference directly in the space of raw sensory representations, such as pixel values of images. Representation learning allows us to automatically discover suitable representations from raw sensory data. For example, given raw sensory data, a deep neural network learns nonlinear representations at its hidden layers, which are subsequently used for classification (or regression) at its output layer. This happens implicitly during training through minimizing a supervised or unsupervised loss in common practical regimes of deep learning, unlike the neural tangent kernel (NTK) regime. In this paper, we study the dynamics of such implicit nonlinear representation learning, which is beyond the NTK regime. We identify a pair of a new assumption and a novel condition, called the common model structure assumption and the data-architecture alignment condition. Under the common model structure assumption, the data-architecture alignment condition is shown to be sufficient for the global convergence and necessary for the global optimality. Moreover, our theory explains how and when increasing the network size does and does not improve the training behaviors in the practical regime. Our results provide practical guidance for designing a model structure: e.g., the common model structure assumption can be used as a justification for using a particular model structure instead of others. We also derive a new training framework based on the theory. The proposed framework is empirically shown to maintain competitive (practical) test performances while providing global convergence guarantees for deep residual neural networks with convolutions, skip connections, and batch normalization with standard benchmark datasets, including CIFAR-10, CIFAR-100, and SVHN.
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Optimization and Control (math.OC); Machine Learning (stat.ML)
Journal reference: Neural computation, volume 34, pages 991-1018 (2022)
DOI: 10.1162/neco_a_01483
Cite as: arXiv:2106.14836 [cs.LG]
  (or arXiv:2106.14836v4 [cs.LG] for this version)

Submission history

From: Kenji Kawaguchi [view email]
[v1] Mon, 28 Jun 2021 16:31:30 GMT (418kb,D)
[v2] Wed, 6 Apr 2022 06:07:38 GMT (413kb,D)
[v3] Thu, 7 Apr 2022 03:36:10 GMT (413kb,D)
[v4] Sat, 9 Apr 2022 05:54:02 GMT (413kb,D)

Link back to: arXiv, form interface, contact.