PHMM
Pair hidden Markov model, a variant of the hidden Markov model designed for sequence alignment that models correspondences between elements of two sequences through paired states; it is commonly used for sequence alignment in bioinformatics.
Timeline (last 90 days)
A recent study followed the CRISP-DM process and used a paired hidden Markov model (PHMM) to align individual RF detections with AIS tracks.
The paired hidden Markov model (PHMM) is a variant of the hidden Markov model (HMM) designed specifically for sequence alignment, modeling two sequences simultaneously and introducing paired states to represent correspondences.
PHMMs are commonly used in bioinformatics for protein or DNA sequence alignment.
On the validation set with multiple candidate detections, the length-corrected Forward score reached a Top-1 precision of 0.7776.
Among detections that passed pre-filtering, the AUC for detecting "dark ships" was 0.637.
After introducing the log-odds score, precision rose to 0.8580 and AUC rose to 0.852.
The study shows that the single-detection, hand-designed PHMM formulation approaches but fails to surpass simple baselines.
Future work may require more complex models or more RF detection information to improve performance.
All Facts (8)
A recent study followed the CRISP-DM process and used a paired hidden Markov model (PHMM) to align individual RF detections with AIS tracks.
50%UnverifiedThe paired hidden Markov model (PHMM) is a variant of the hidden Markov model (HMM) designed specifically for sequence alignment, modeling two sequences simultaneously and introducing paired states to represent correspondences.
50%UnverifiedPHMMs are commonly used in bioinformatics for protein or DNA sequence alignment.
50%UnverifiedOn the validation set with multiple candidate detections, the length-corrected Forward score reached a Top-1 precision of 0.7776.
50%UnverifiedAmong detections that passed pre-filtering, the AUC for detecting "dark ships" was 0.637.
50%UnverifiedAfter introducing the log-odds score, precision rose to 0.8580 and AUC rose to 0.852.
50%UnverifiedThe study shows that the single-detection, hand-designed PHMM formulation approaches but fails to surpass simple baselines.
50%UnverifiedFuture work may require more complex models or more RF detection information to improve performance.
50%