Diffeomorphic Temporal Alignment Nets
Authors: Ron A. Shapira Weber, Matan Eyal, Nicki Skafte, Oren Shriki, Oren Freifeld
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluated DTAN s time-series joint alignment of both synthetic and real-world data. |
| Researcher Affiliation | Academia | Ron Shapira Weber Ben-Gurion University ronsha@post.bgu.ac.il; Matan Eyal Ben-Gurion University mataney@post.bgu.ac.il; Nicki Skafte Detlefsen Technical University of Denmark nsde@dtu.dk; Oren Shriki Ben-Gurion University shrikio@bgu.ac.il; Oren Freifeld Ben-Gurion University orenfr@cs.bgu.ac.il |
| Pseudocode | No | The paper describes its methods in narrative text and mathematical equations, but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Our code is available at https://github.com/BGU-CS-VIL/dtan. |
| Open Datasets | Yes | The UCR time-series classification archive [8] contains 85 real-world datasets (we used 84). ... www.cs.ucr.edu/~eamonn/time_series_data/. |
| Dataset Splits | Yes | We generated 250 samples per-class (1000 in total) and used a 60-20-20% train, validation and test split, choosing the model with the lowest validation loss |
| Hardware Specification | Yes | Timing was measured on a Nvidia Ge Force GTX 1080 graphic card. |
| Software Dependencies | No | The paper mentions 'Tensorflow C++ API, and Keras wrapper' and 'Tensorflow [1] or Keras [9]' but does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | For simplicity, in our experiments floc is set to be a 1D CNN consisting of 3 conv-layers (128 64 64 filters per layer, respectively) each followed by a Re LU nonlinear activation function [40], batch-normalization and max-pooling layers [27], where d = dim(θ) = 32. The learning rate was η = 10 4, set to minimize Eq. (6) via the Adam optimizer [30]. The last activation function was tanh. ... where λvar [10 3, 10 2], λsmooth [0.5, 1]. We used R-DTANx, where x {1, 2, 4} is the number of TT layers. |