🚀 Interested in aligning time series data?
This week at
#ICASSP2025, Afek Steinberg presents our new method:
🌀 Conditional Deep Canonical Time Warping (CDCTW) — a new method for aligning high-dim sequences (think speech, video, biosignals).
✨ Main idea:
✅ A new dynamic deep Canonical Correlation Analysis for learning a shared embedding
✅ Conditional stochastic feature selection attenuates noise
✅ Warping signals in the shared embedding leads to SOTA results on real-world benchmarks
👏 Amazing work by Afek & Ran
📍Come for more details at
@ieeeICASSP
📅 April 11, 14:00 | ML for Time Series 1
📄 Paper:
arxiv.org/abs/2412.18234
#MachineLearning #TimeSeries #DeepLearning #TemporalAlignment #FeatureSelection