Two journal papers have been accepted in the last few weeks:

• A.Băltoiu, Denis C. Ilie-Ablachim, B.Dumitrescu, “Dictionary learning with cone atoms and application to anomaly detection”, Signal Processing
• Denis C. Ilie-Ablachim, A.Băltoiu, B.Dumitrescu, “Sparse representation with Gaussian atoms and its use in anomaly detection”, IEEE Open Journal of Signal Processing

They both contain algorithms for sparse representation and dictionary learning using atoms that are sets instead of vectors. For some classes of anomalies, they obtain better than state-of-the-art detection.

Two papers have been accepted to IEEE International Workshop on Machine Learning in Signal Processing (Rome, Sept 2023):

• D.C. Ilie-Ablachim, B.Dumitrescu, “Classification with dictionary learning and a distance barrier promoting incoherence”
• C.E. Zica, B.Dumitrescu, “Online Computation of Reduced Egonet Features for Anomaly Detection in Bank Transactions Graphs”

Paper “Incoherent frames design and dictionary learning using a distance barrier” accepted in Signal Processing journal. For many medium and large frame sizes, the proposed algorithm gives the best mutual coherence.

Paper “Sparse optimization with cone atoms” accepted at ICASSP 2023!

New programs for incoherent frame design, with some of the best ever results! Also, a related method for dictionary learning. Check the Software and Publications sections.

We have started to work on the project, but this site is still under construction. Please use with caution.