Update paper.md

paper/paper.md

@@ -36,9 +36,8 @@ Adaptive Brownian bridge-based aggregation (ABBA) [@EG19b] is a symbolic time se
 ABBA follows a two-phase approach to symbolize time series, namely compression and digitization. The first phase aims to reduce the time series dimension by polygonal chain approximation, and the second phase assigns symbols to the polygonal pieces. Both phases operate together to ensure that the essential time series features are best reflected by the symbols, controlled by a user-chosen error tolerance. The advantages of the ABBA representation against other symbolic representations include (1) better preservation of essential shape features, e.g., when compared against the popular SAX representation~[@SAX03; @EG19b]; (2) effective representation of local up and down trends in the time series which supports motif detection; (3) demonstrably reduced sensitivity to hyperparameters of neural network models and the initialization of random weights in forecasting applications~[@EG20b].  
 
 
-*fABBA* is a Python library to compute ABBA symbolic time series representations on Linux, Windows, and MacOS systems. With Cython compilation and typed memoryviews, it significantly outperforms existing ABBA implementations. The *fABBA* library also includes a new ABBA variant, fABBA [@CG22a], which uses an alternative digitization method (``greedy aggregation'') instead of k-means clustering~[@1056489], providing significant speedup and improved tolerance-based digitization (without the need to specify the number $k$ of symbols a priori). The experiments in [@CG22a] demonstrate that fABBA compares favorably to the original ABBA module[^1] in terms of runtime. *fABBA* is an open-source library and licensed under the 3-Clause BSD License. Its redistribution and use, with or without modification, are permitted under conditions described in `https://opensource.org/license/bsd-3-clause/`.
+*fABBA* is a Python library to compute ABBA symbolic time series representations on Linux, Windows, and MacOS systems. With Cython compilation and typed memoryviews, it significantly outperforms existing ABBA implementations. The *fABBA* library also includes a new ABBA variant, fABBA [@CG22a], which uses an alternative digitization method (``greedy aggregation'') instead of k-means clustering~[@1056489], providing significant speedup and improved tolerance-based digitization (without the need to specify the number $k$ of symbols a priori). The experiments in [@CG22a] demonstrate that fABBA runs significantly faster than the original ABBA module at `https://github.com/nla-group/ABBA/`. *fABBA* is an open-source library and licensed under the 3-Clause BSD License. Its redistribution and use, with or without modification, are permitted under conditions described in `https://opensource.org/license/bsd-3-clause/`.
 
-[^1] https://github.com/nla-group/ABBA/
 
 # Examples
 *fABBA*  can installed via the Python Package Index or conda forge. Detailed documentation for its installation, usage, API reference, and quick start examples can be found on~`https://fabba.readthedocs.io/en/latest/`. Below we provide a brief demonstration.