• Corpus ID: 211677719

Unsupervised Dictionary Learning for Anomaly Detection

@article{Irofti2020UnsupervisedDL,
  title={Unsupervised Dictionary Learning for Anomaly Detection},
  author={Paul Irofti and Andra Baltoiu},
  journal={ArXiv},
  year={2020},
  volume={abs/2003.00293}
}
We investigate the possibilities of employing dictionary learning to address the requirements of most anomaly detection applications, such as absence of supervision, online formulations, low false positive rates. We present new results of our recent semi-supervised online algorithm, TODDLeR, on a anti-money laundering application. We also introduce a novel unsupervised method of using the performance of the learning algorithm as indication of the nature of the samples. 

Figures from this paper

Dictionary Learning with Uniform Sparse Representations for Anomaly Detection
TLDR
This paper uses a particular DL formulation that seeks uniform sparse representations model to detect the underlying subspace of the majority of samples in a dataset, using a K-SVD-type algorithm.
Fraud Detection in Networks
TLDR
A survey on anomaly detection techniques used for fraud detection that exploit both the graph structure underlying the data and the contextual information contained in the attributes that are suited for financial frauds.
Fraud Detection in Networks: State-of-the-art
TLDR
A survey on anomaly detection techniques used for fraud detection that exploit both the graph structure underlying the data and the contextual information contained in the attributes are presented.

References

SHOWING 1-10 OF 14 REFERENCES
Learning a discriminative dictionary for sparse coding via label consistent K-SVD
TLDR
A label consistent K-SVD (LC-KSVD) algorithm to learn a discriminative dictionary for sparse coding is presented, which learns a single over-complete dictionary and an optimal linear classifier jointly and yields dictionaries so that feature points with the same class labels have similar sparse codes.
Recursive least squares dictionary learning
  • IEEE Trans. Signal Proc.,
  • 2010
Malware Identification with Dictionary Learning
TLDR
A low complexity semi-supervised dictionary learning framework that begins with training an initial dictionary on a small labeled data set, and then continues with online learning on incoming unlabeled data, with the scope of adapting to new and unknown malware types.
Fraud Detection in Networks: State-of-the-art
TLDR
A survey on anomaly detection techniques used for fraud detection that exploit both the graph structure underlying the data and the contextual information contained in the attributes are presented.
Community-Level Anomaly Detection for Anti-Money Laundering
TLDR
This work investigates the problem of learning graph structure representations using adaptations of dictionary learning aimed at encoding connectivity patterns and proposes new methods that impose Laplacian structure on the dictionaries themselves.
Dictionary Learning Algorithms and Applications
Method of optimal directions for frame design
  • K. Engan, S. O. Aase, J. H. Husøy
  • Computer Science
    1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258)
  • 1999
TLDR
Experiments demonstrate that the approximation capabilities, in terms of mean squared error (MSE), of the optimized frames are significantly better than those obtained using frames designed by the algorithm of Engan et.
Recursive Least Squares Dictionary Learning Algorithm
TLDR
The recursive least squares dictionary learning algorithm, RLS-DLA, is presented, which can be used for learning overcomplete dictionaries for sparse signal representation and a forgetting factor can be introduced and easily implemented in the algorithm.
Label consistent recursive least squares dictionary learning for image classification
TLDR
A label consistent recursive least squares dictionary learning algorithm, LC-RLSDLA, to learn discriminative dictionaries for image classification based on sparse coding outperforms state-of-the-art sparse coding algorithms.
...
...