Tweets by @bendfulcher about time-series analysis.
Sydney, New South Wales
Joined July 2015
- Tweets492
- Following79
- Followers764
- Likes267
Time Series Features retweeted
What kinds of dynamical structures have scientists developed models for?
Introducing the 1000×1000 collection: 1000 simulated time series × 1000 samples from 133 dynamical processes, from noise and chaos to hearts, brains and climate.
Have a play? 🐛
dynamicsandneuralsystems.git…
Time Series Features retweeted
Rivers, pendulums and the thalamus: how the brain balances stable and flexible dynamics, and what changes when consciousness fades.
Our latest journey into how the thalamus is shaping cortical dynamics in @ScienceAdvances 🧵1/10
science.org/doi/10.1126/scia…
Time Series Features retweeted
New preprint!: "Using matrix-product states for time-series machine learning".
arxiv.org/abs/2412.15826
Quick summary below 👇
Time Series Features retweeted
Better believe it, there are now TWO #timeseries feature sets available in #julialang.
The new CatchaMouse16.jl package joins Catch22.jl, bringing 16 more features tailored to (mouse) fMRI data: github.com/brendanjohnharris…
Check out the CatchaMouse16 paper below
New preprint w/ Imran Alam, Patrick Cahill @Valerio_Zerbi @m_markicevic @brendanjohnh @olivercliff
"Canonical time-series features for characterizing biologically informative dynamical patterns in fMRI"
biorxiv.org/content/10.1101/…
Code: github.com/DynamicsAndNeural…
Short summary 👇
Time Series Features retweeted
A new method of detecting criticality from time-series data outperforms conventional metrics in the presence of variable noise levels for both simulated systems and real neural recordings.
Read go.aps.org/3WBjcXf
#PRXjustpublished #PRXopenaccess #PRXComplexSystems
Time Series Features retweeted
Our work by @brendanjohnh (w Leo Gollo) on tracking the distance to criticality in noisy systems is now out in @PhysRevX 🙂 (includes an application tracking criticality across the mouse visual hierarchy)
doi.org/10.1103/PhysRevX.14.…
Code details: time-series-features.gitbook…
A new method of detecting criticality from time-series data outperforms conventional metrics in the presence of variable noise levels for both simulated systems and real neural recordings.
Read go.aps.org/3WBjcXf
#PRXjustpublished #PRXopenaccess #PRXComplexSystems
Time Series Features retweeted
New preprint by Rishi Maran @eli_j_muller
"Analyzing the Brain's Dynamic Response to Targeted Stimulation using Generative Modeling"
A review/perspective on why new mechanisms may be found by modeling brain stimulation dynamics 🧠⚡️
arxiv.org/abs/2407.19737
Quick summary 👇
Analyzing the Brain's Dynamic Response to Targeted Stimulation using Generative Modeling. arxiv.org/abs/2407.19737
Time Series Features retweeted
New preprint w/ Imran Alam, Patrick Cahill @Valerio_Zerbi @m_markicevic @brendanjohnh @olivercliff
"Canonical time-series features for characterizing biologically informative dynamical patterns in fMRI"
biorxiv.org/content/10.1101/…
Code: github.com/DynamicsAndNeural…
Short summary 👇
Time Series Features retweeted
Latest preprint: "Parameter Inference from a Non-stationary Unknown Process" (PINUP)
We're really interested in the problem of inferring sources of non-stationary variation directly from measured time-series data.
arxiv.org/abs/2407.08987v1
Quick summary 👇
Time Series Features retweeted
If you're at OHBM this year, check out @AnnieGBryant's great work developing a systematic method to extract interpretable dynamical patterns from fMRI time series!
#OHBM very excited to share this (v2.0) at the 'Transdiagnostic Perspectives on Neurodevelopmental and Psychiatric Disorders - Part 1' 12pm oral session on Tuesday, and poster #1740 on Wed/Thurs afternoon! Come say hi 😊
Time Series Features retweeted
Amazing work using the great pyspi package! We also used it in our recent work and examined the sensitivity of (only) 20 representative FC metrics regarding neural decline induced by age and malignant brain tumors doi.org/10.1101/2024.03.18.5…!
Time Series Features retweeted
Benchmarking methods for mapping functional connectivity in the brain | doi.org/10.1101/2024.05.07.5…
What is the best FC metric?
Led by @liuzhenqi0303 avec @loopyluppi @JustineYHansen @yetianmed @AndrewZalesky @bttyeo @bendfulcher ⤵️
Time Series Features retweeted
#Satellite ComplexTime explores temporal dynamics in complex systems across various domains. They invite submissions on topics related to temporal data handling, methods, and tools. Learn more at: sites.google.com/view/comple…
Time Series Features retweeted
bake off redux is online first, it reviews and compares recent algorithms for time series classification using the @aeon_toolkit for the vast majority of experiments link.springer.com/article/10…
Time Series Features retweeted
New results now live! Most exciting is @brendanjohnh mouse #neuropixels application.
We find that brain areas higher in the visual hierarchy are closer to #criticality (in a way that cannot be detected with existing #timeseries measures) 😄🐭⬇️
arxiv.org/abs/2310.14791
New #ComplexSystems preprint:
"Tracking the distance to criticality in systems with unknown noise"
By @brendanjohnh w/ Leonardo Gollo 😀
arxiv.org/abs/2310.14791
A summary in the thread below 👇
Time Series Features retweeted
Looking at training of a neural network as a (temporal) network trajectory, we can investigate the blackbox of machine learning training through dynamical systems and network theory.
Great & fun collab w/ @kaloyandanovski & @miguelcsoriano
preprint:
arxiv.org/pdf/2404.05782.pdf
Time Series Features retweeted
New pre-print by @aria_mt_nguyen w @jlizier:
"A feature-based information-theoretic approach for detecting interpretable, long-timescale pairwise interactions from time series"
Introduces a new method: uses features to infer time-series interactions
arxiv.org/abs/2404.05929
Time Series Features retweeted
Reminder that I'll be running a second edition of my Ecological forecasting in R workshop in May this year. Learn to use #mgcv, #mvgam and #brms to analyse complex ecological time series and produce meaningful insights #rstats physalia-courses.org/courses…
Time Series Features retweeted
We released aeon v0.8, which adds more algorithms for time series regression, along with some deprecation, bugfixes and visualisation function. You can now also use unequal length series for MSM and SBD pairwise distance ! Full changelog available on GitHub.
Time Series Features retweeted
We talked with @azulgarza_ and @MergenthalerMax.
They are co-founders of @nixtlainc and researchers behind TimeGPT, the first foundation model designed specifically for time series forecasting!
Read our interview here: turingpost.com/p/timegpt
