Adaptive Loss Minimization for Semi-Supervised Elastic Embedding

نویسندگان

  • Feiping Nie
  • Hua Wang
  • Heng Huang
  • Chris H. Q. Ding
چکیده

The semi-supervised learning usually only predict labels for unlabeled data appearing in training data, and cannot effectively predict labels for testing data never appearing in training set. To handle this outof-sample problem, many inductive methods make a constraint such that the predicted label matrix should be exactly equal to a linear model. In practice, this constraint is too rigid to capture the manifold structure of data. Motivated by this deficiency, we relax the rigid linear embedding constraint and propose to use an elastic embedding constraint on the predicted label matrix such that the manifold structure can be better explored. To solve our new objective and also a more general optimization problem, we study a novel adaptive loss with efficient optimization algorithm. Our new adaptive loss minimization method takes the advantages of both L1 norm and L2 norm, and is robust to the data outlier under Laplacian distribution and can efficiently learn the normal data under Gaussian distribution. Experiments have been performed on image classification tasks and our approach outperforms other state-of-the-art methods.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Combination of Adaptive-Grid Embedding and Redistribution Methods on Semi Structured Grids for two-dimensional invisid flows

Among the adaptive-grid methods, redistribution and embedding techniques have been the focus of more attention by researchers. Simultaneous or combined adaptive techniques have also been used. This paper describes a combination of adaptive-grid embedding and redistribution methods on semi-structured grids for two-dimensional invisid flows. Since the grid is semi-structured, it is possible to us...

متن کامل

Combination of Adaptive-Grid Embedding and Redistribution Methods on Semi Structured Grids for two-dimensional invisid flows

Among the adaptive-grid methods, redistribution and embedding techniques have been the focus of more attention by researchers. Simultaneous or combined adaptive techniques have also been used. This paper describes a combination of adaptive-grid embedding and redistribution methods on semi-structured grids for two-dimensional invisid flows. Since the grid is semi-structured, it is possible to us...

متن کامل

Effects of Directional Subdividing on adaptive Grid-Embedding (RESEARCH NOTE)

The effects of using both directions and directional subdividing on adaptive gridembedding on the computational time and the number of grid points required for the same accuracy are considered. Directional subdividing is used from the beginning of the adaptation procedure without any restriction. To avoid the complication of unstructured grid, the semi-structured grid was used. It is used to so...

متن کامل

Dimensionality Reduction by Supervised Neighbor Embedding Using Laplacian Search

Dimensionality reduction is an important issue for numerous applications including biomedical images analysis and living system analysis. Neighbor embedding, those representing the global and local structure as well as dealing with multiple manifolds, such as the elastic embedding techniques, can go beyond traditional dimensionality reduction methods and find better optima. Nevertheless, existi...

متن کامل

Semi-supervised Graph Embedding Approach to Dynamic Link Prediction

We propose a simple discrete time semi–supervised graph embedding approach to link prediction in dynamic networks. The learned embedding reflects information from both the temporal and cross–sectional network structures, which is performed by defining the loss function as a weighted sum of the supervised loss from past dynamics and the unsupervised loss of predicting the neighborhood context in...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2013