it's crazy the optimizations you can make when you don't care about the least significant bit or two? (if you're quantizing to 16-bit audio, LSB is below quant noise threshold)
converting bit-perfect resampling to use a polyphase table alone was like 64x faster
converting bit-perfect resampling to use a polyphase table alone was like 64x faster
November 15, 2026 at 2:57 PM
it's crazy the optimizations you can make when you don't care about the least significant bit or two? (if you're quantizing to 16-bit audio, LSB is below quant noise threshold)
converting bit-perfect resampling to use a polyphase table alone was like 64x faster
converting bit-perfect resampling to use a polyphase table alone was like 64x faster
The Move Pixels tool now has a "Nearest Neighbor (Antialiased)" resampling mode. Similar to what retro game emulators call "bilinear sharp." It's NN but with antialiased edges:
(sprite credit: Stardew Valley
x.com/ConcernedApe)
(sprite credit: Stardew Valley
x.com/ConcernedApe)
November 10, 2026 at 8:52 PM
The Move Pixels tool now has a "Nearest Neighbor (Antialiased)" resampling mode. Similar to what retro game emulators call "bilinear sharp." It's NN but with antialiased edges:
(sprite credit: Stardew Valley
x.com/ConcernedApe)
(sprite credit: Stardew Valley
x.com/ConcernedApe)
Finally got a #Raspberry Pi #RP2040 Needed something with I2S support to test a #MAXIM #MAX98357A I2S 3W Class D #Amplifier Breakout board.
Still took me most of day trouble shooting wiring, code, and resampling a wav file to get it to work
Still took me most of day trouble shooting wiring, code, and resampling a wav file to get it to work
November 14, 2026 at 6:43 AM
Finally got a #Raspberry Pi #RP2040 Needed something with I2S support to test a #MAXIM #MAX98357A I2S 3W Class D #Amplifier Breakout board.
Still took me most of day trouble shooting wiring, code, and resampling a wav file to get it to work
Still took me most of day trouble shooting wiring, code, and resampling a wav file to get it to work
Do you remember an N, Miranda?
Do you remember an N?
And the sampling and exampling
With the for() loop resampling
And the "p"s that tease
With their "t"s and their "z"s
And the code that tasted of R?
Do you remember an N?
And the sampling and exampling
With the for() loop resampling
And the "p"s that tease
With their "t"s and their "z"s
And the code that tasted of R?
November 11, 2026 at 8:44 AM
Do you remember an N, Miranda?
Do you remember an N?
And the sampling and exampling
With the for() loop resampling
And the "p"s that tease
With their "t"s and their "z"s
And the code that tasted of R?
Do you remember an N?
And the sampling and exampling
With the for() loop resampling
And the "p"s that tease
With their "t"s and their "z"s
And the code that tasted of R?
The ASR-10 converters give samples a lovely grit. Resampling a loop back through it once is a nice way to thicken things without reaching for another effect.
November 14, 2026 at 1:01 AM
The ASR-10 converters give samples a lovely grit. Resampling a loop back through it once is a nice way to thicken things without reaching for another effect.
Numbers come from the per-session jsonl in landmine's repo (commit c6d0123), Sonnet 5.5 with shell, 6 sessions per task. Resampling tasks gives a gain of +14 points, 95% CI +3 to +28.
November 15, 2026 at 6:45 PM
Numbers come from the per-session jsonl in landmine's repo (commit c6d0123), Sonnet 5.5 with shell, 6 sessions per task. Resampling tasks gives a gain of +14 points, 95% CI +3 to +28.
Where do your sounds come from? Sample packs, your own synths, field recordings, or resampling your own stuff? Tell us your secret source 👇
November 12, 2026 at 9:08 AM
Where do your sounds come from? Sample packs, your own synths, field recordings, or resampling your own stuff? Tell us your secret source 👇
Curious how you run the Monte Carlo: resampling your own trades, or generating fresh price paths? Resampling trades shows how rough the ride could get, but not whether the rules were fit to that data. That gap is why my Lab posts its fails as loudly as its passes.
November 11, 2026 at 9:34 AM
Curious how you run the Monte Carlo: resampling your own trades, or generating fresh price paths? Resampling trades shows how rough the ride could get, but not whether the rules were fit to that data. That gap is why my Lab posts its fails as loudly as its passes.
New on CRAN: imbalanceKit (0.1.0). View at https://CRAN.R-project.org/package=imbalanceKit
imbalanceKit: Resampling Methods and Tools for Imbalanced Classification
A collection of resampling methods for class imbalance in classification problems. Includes the oversampling methods SMOTE with boosting (SMOTEWB) by F. Sağlam and M. A. Cengiz (2022) <<a href="https://doi.org/10.1016%2Fj.eswa.2022.117023" target="_top">doi:10.1016/j.eswa.2022.117023</a>>, Markov chain SMOTE (MCSMOTE) by F. Sağlam and M. A. Cengiz (2025) <<a href="https://doi.org/10.1016%2Fj.asoc.2025.113906" target="_top">doi:10.1016/j.asoc.2025.113906</a>>, uncertainty guided oversampling (UGO) by F. Sağlam and M. A. Cengiz (2026) <<a href="https://doi.org/10.1016%2Fj.knosys.2026.115886" target="_top">doi:10.1016/j.knosys.2026.115886</a>>, and bi-directional SMOTE (BiSMOTE) by F. Sağlam and U. Yamak (2027) <<a href="https://doi.org/10.1016%2Fj.eswa.2026.134648" target="_top">doi:10.1016/j.eswa.2026.134648</a>>, and the data relocation approach DatRel by F. Sağlam (2025) <<a href="https://doi.org/10.1007%2Fs10994-025-06755-8" target="_top">doi:10.1007/s10994-025-06755-8</a>>, together with commonly used alternatives such as SMOTE, ADASYN, Borderline-SMOTE, Safe-Level-SMOTE, geometric SMOTE, random oversampling and undersampling. The methods share the same interface: they take a feature matrix and a factor class variable and return the resampled data together with the synthetic samples.
CRAN.R-project.org
November 13, 2026 at 8:12 PM
New on CRAN: imbalanceKit (0.1.0). View at https://CRAN.R-project.org/package=imbalanceKit
3/4Algoritmo
Inicialização: Joga-se uma nuvem de partículas aleatórias. Predição: O modelo físico avança as partículas no tempo. Atualização (Peso): Cada partícula recebe um peso conforme sua proximidade com a realidade medida. Resampling: As ruins somem, as boas se clonam. Evolução darwiniana pura!
Inicialização: Joga-se uma nuvem de partículas aleatórias. Predição: O modelo físico avança as partículas no tempo. Atualização (Peso): Cada partícula recebe um peso conforme sua proximidade com a realidade medida. Resampling: As ruins somem, as boas se clonam. Evolução darwiniana pura!
November 14, 2026 at 10:56 PM
3/4Algoritmo
Inicialização: Joga-se uma nuvem de partículas aleatórias. Predição: O modelo físico avança as partículas no tempo. Atualização (Peso): Cada partícula recebe um peso conforme sua proximidade com a realidade medida. Resampling: As ruins somem, as boas se clonam. Evolução darwiniana pura!
Inicialização: Joga-se uma nuvem de partículas aleatórias. Predição: O modelo físico avança as partículas no tempo. Atualização (Peso): Cada partícula recebe um peso conforme sua proximidade com a realidade medida. Resampling: As ruins somem, as boas se clonam. Evolução darwiniana pura!
Modem update is live! Turns out resampling is super cheap when operating in the oversampled time domain- Lagrange interpolation is all you need. Refresh the page to get it.
November 11, 2026 at 5:03 PM
Modem update is live! Turns out resampling is super cheap when operating in the oversampled time domain- Lagrange interpolation is all you need. Refresh the page to get it.
it's pretty new so support / longevity be good and playing with it gave me serious gadget envy, it's got some real nice features my sp-404 mkII doesn't have. resampling is a lil clunky. but ohh, what a screen. a well designed UI is sooo good to have. seems a v good bang for buck, i say look into it!
November 11, 2026 at 5:37 PM
it's pretty new so support / longevity be good and playing with it gave me serious gadget envy, it's got some real nice features my sp-404 mkII doesn't have. resampling is a lil clunky. but ohh, what a screen. a well designed UI is sooo good to have. seems a v good bang for buck, i say look into it!
Yeah, I hear that. AR and Digitone MKI are on the chopping block. M:C is gone.
Do you use samples on the AR? I remember having lots of fun with them.
MKII at least has resampling, which opens up lots of possibilities.
Do you use samples on the AR? I remember having lots of fun with them.
MKII at least has resampling, which opens up lots of possibilities.
November 11, 2026 at 4:14 PM
Yeah, I hear that. AR and Digitone MKI are on the chopping block. M:C is gone.
Do you use samples on the AR? I remember having lots of fun with them.
MKII at least has resampling, which opens up lots of possibilities.
Do you use samples on the AR? I remember having lots of fun with them.
MKII at least has resampling, which opens up lots of possibilities.
The biggest bottleneck in GeoAI isn't the model, it's the pipeline.
Satellite imagery, radar, DEMs, land-cover products, QA bands, cloud masks, reprojection, resampling, harmonization... time gets lost.
This paper focuses on solving that problem, and I think it's worth paying attention to.
Satellite imagery, radar, DEMs, land-cover products, QA bands, cloud masks, reprojection, resampling, harmonization... time gets lost.
This paper focuses on solving that problem, and I think it's worth paying attention to.
GeoAI Imagery Data Pipelines
The Earth observation community is living through a golden age of AI. Today, it's easier than ever to download a pretrained model, fine tune it on satellite...
davidgsmith.net
November 15, 2026 at 3:21 PM
The biggest bottleneck in GeoAI isn't the model, it's the pipeline.
Satellite imagery, radar, DEMs, land-cover products, QA bands, cloud masks, reprojection, resampling, harmonization... time gets lost.
This paper focuses on solving that problem, and I think it's worth paying attention to.
Satellite imagery, radar, DEMs, land-cover products, QA bands, cloud masks, reprojection, resampling, harmonization... time gets lost.
This paper focuses on solving that problem, and I think it's worth paying attention to.
Want to see cost sensitive learning done right?
This scikit learn example tunes the decision threshold to minimise business cost on credit and fraud data, with no resampling.
Worth bookmarking 👇
This scikit learn example tunes the decision threshold to minimise business cost on credit and fraud data, with no resampling.
Worth bookmarking 👇
Post-tuning the decision threshold for cost-sensitive learning
Once a classifier is trained, the output of the predict method outputs class label predictions corresponding to a thresholding of either the decision_function or the predict_proba output. For a bin...
scikit-learn.org
November 13, 2026 at 4:00 PM
Want to see cost sensitive learning done right?
This scikit learn example tunes the decision threshold to minimise business cost on credit and fraud data, with no resampling.
Worth bookmarking 👇
This scikit learn example tunes the decision threshold to minimise business cost on credit and fraud data, with no resampling.
Worth bookmarking 👇
NEW: GPUResize 2. Fast, GPU-accelerated resizing with 22 resampling methods. New in v2: Free Distort, Mosaic, post-sharpening, interpolation spaces and DaVinci Resolve (OFX) support.
aescripts.com/gpuresize/
#aftereffects #davinciresolve #aescripts
aescripts.com/gpuresize/
#aftereffects #davinciresolve #aescripts
November 9, 2026 at 7:08 PM
NEW: GPUResize 2. Fast, GPU-accelerated resizing with 22 resampling methods. New in v2: Free Distort, Mosaic, post-sharpening, interpolation spaces and DaVinci Resolve (OFX) support.
aescripts.com/gpuresize/
#aftereffects #davinciresolve #aescripts
aescripts.com/gpuresize/
#aftereffects #davinciresolve #aescripts
I prefer the random cycling of the residuals fwiw; here are 10 replicates vs the actual time series (L). Pretty realistic counterfactual imo. Just resampling residuals looks too noisy (R)
November 6, 2026 at 6:45 PM
I prefer the random cycling of the residuals fwiw; here are 10 replicates vs the actual time series (L). Pretty realistic counterfactual imo. Just resampling residuals looks too noisy (R)
New CRAN package imbalanceKit with initial version 0.1.0
#rstats
https://cran.r-project.org/package=imbalanceKit
#rstats
https://cran.r-project.org/package=imbalanceKit
CRAN: Package imbalanceKit
A collection of resampling methods for class imbalance in classification problems. Includes the oversampling methods SMOTE with boosting (SMOTEWB) by F. Sağlam and M. A. Cengiz (2022) <<a href="https://doi.org/10.1016%2Fj.eswa.2022.117023" target="_top">doi:10.1016/j.eswa.2022.117023</a>>, Markov chain SMOTE (MCSMOTE) by F. Sağlam and M. A. Cengiz (2025) <<a href="https://doi.org/10.1016%2Fj.asoc.2025.113906" target="_top">doi:10.1016/j.asoc.2025.113906</a>>, uncertainty guided oversampling (UGO) by F. Sağlam and M. A. Cengiz (2026) <<a href="https://doi.org/10.1016%2Fj.knosys.2026.115886" target="_top">doi:10.1016/j.knosys.2026.115886</a>>, and bi-directional SMOTE (BiSMOTE) by F. Sağlam and U. Yamak (2027) <<a href="https://doi.org/10.1016%2Fj.eswa.2026.134648" target="_top">doi:10.1016/j.eswa.2026.134648</a>>, and the data relocation approach DatRel by F. Sağlam (2025) <<a href="https://doi.org/10.1007%2Fs10994-025-06755-8" target="_top">doi:10.1007/s10994-025-06755-8</a>>, together with commonly used alternatives such as SMOTE, ADASYN, Borderline-SMOTE, Safe-Level-SMOTE, geometric SMOTE, random oversampling and undersampling. The methods share the same interface: they take a feature matrix and a factor class variable and return the resampled data together with the synthetic samples.
cran.r-project.org
November 13, 2026 at 5:02 PM
New CRAN package imbalanceKit with initial version 0.1.0
#rstats
https://cran.r-project.org/package=imbalanceKit
#rstats
https://cran.r-project.org/package=imbalanceKit
Resampling matched weeks puts this year's rise at 1.5% to 4.0%; it describes how much the comparison of the two years' weeks could vary, not the accuracy of each count.
November 8, 2026 at 2:00 PM
Resampling matched weeks puts this year's rise at 1.5% to 4.0%; it describes how much the comparison of the two years' weeks could vary, not the accuracy of each count.
StepBack: Step-Level Error-Localized Resampling for Efficient Test-Time Reasoning
Yang Ouyang, Jung-Eun Kim
#arXiv #cs.AI #cs.CL #cs.LG
Yang Ouyang, Jung-Eun Kim
#arXiv #cs.AI #cs.CL #cs.LG
StepBack: Step-Level Error-Localized Resampling for Efficient Test-Time Reasoning
Self-consistency (SC) is a widely used decoding strategy for improving chain-of-thought reasoning, but it requires sampling many complete reasoning trajectories with a preset budget. Its efficient variants reduce the cost mainly by stopping full-trajectory sampling early or distributing the number …
arxiv.org
November 7, 2026 at 2:45 AM
This was a nice paper, but it was unfortunate that it didn’t mention the large, relevant stats literature on other resampling methods that address the same issues. For example, it doesn’t mention Fay’s BRR method, which is essentially a weighted bootstrap long used at eg the US Census Bureau.
November 7, 2026 at 1:40 PM
This was a nice paper, but it was unfortunate that it didn’t mention the large, relevant stats literature on other resampling methods that address the same issues. For example, it doesn’t mention Fay’s BRR method, which is essentially a weighted bootstrap long used at eg the US Census Bureau.
[今年買ったプラグイン]
Phil Speiser の THE_Pattern 2。
Presetが多彩で秀逸。欲しい音とパターンが最初から入ってる。
気に入ったパターンを読み込むと、IntroからOutroまで提案してくれる。手持ちのSampleで簡単にKitを作成、DAWにExport可能。
KickのSampleを読み込みResamplingしてEditできるので、お好みのKickを作れます。こんなドラム音源はいままで無かった。
www.philspeiser.com/product/the_...
Phil Speiser の THE_Pattern 2。
Presetが多彩で秀逸。欲しい音とパターンが最初から入ってる。
気に入ったパターンを読み込むと、IntroからOutroまで提案してくれる。手持ちのSampleで簡単にKitを作成、DAWにExport可能。
KickのSampleを読み込みResamplingしてEditできるので、お好みのKickを作れます。こんなドラム音源はいままで無かった。
www.philspeiser.com/product/the_...
THE_PATTERN 2 – The Drum Machine From The Future
Everything you expect from a drum machine, only smarter. Then the magic: a drum loop rebuilt from your sounds, a beat you beatbox onto the grid, eight bars turned into the whole drum track. Every deci...
www.philspeiser.com
November 12, 2026 at 3:31 PM
[今年買ったプラグイン]
Phil Speiser の THE_Pattern 2。
Presetが多彩で秀逸。欲しい音とパターンが最初から入ってる。
気に入ったパターンを読み込むと、IntroからOutroまで提案してくれる。手持ちのSampleで簡単にKitを作成、DAWにExport可能。
KickのSampleを読み込みResamplingしてEditできるので、お好みのKickを作れます。こんなドラム音源はいままで無かった。
www.philspeiser.com/product/the_...
Phil Speiser の THE_Pattern 2。
Presetが多彩で秀逸。欲しい音とパターンが最初から入ってる。
気に入ったパターンを読み込むと、IntroからOutroまで提案してくれる。手持ちのSampleで簡単にKitを作成、DAWにExport可能。
KickのSampleを読み込みResamplingしてEditできるので、お好みのKickを作れます。こんなドラム音源はいままで無かった。
www.philspeiser.com/product/the_...
bit-perfect says that the system does not alter the music files output - an AIFF 16-Bit 44.1 KHz file just stays the same 16-Bit 44.1 KHz AIFF on its way to the DAC - no extra bass, treble or resampling. HiRes files even sound better.
November 7, 2026 at 7:26 PM
bit-perfect says that the system does not alter the music files output - an AIFF 16-Bit 44.1 KHz file just stays the same 16-Bit 44.1 KHz AIFF on its way to the DAC - no extra bass, treble or resampling. HiRes files even sound better.
Yang Ouyang, Jung-Eun Kim: StepBack: Step-Level Error-Localized Resampling for Efficient Test-Time Reasoning https://arxiv.org/abs/2611.08403 https://arxiv.org/pdf/2611.08403 https://arxiv.org/html/2611.08403
November 6, 2026 at 6:39 AM
Yang Ouyang, Jung-Eun Kim: StepBack: Step-Level Error-Localized Resampling for Efficient Test-Time Reasoning https://arxiv.org/abs/2611.08403 https://arxiv.org/pdf/2611.08403 https://arxiv.org/html/2611.08403
New on CRAN: nutrigon (0.1.0). View at https://CRAN.R-project.org/package=nutrigon
nutrigon: Angular Separation Between Nutritional Optima
Fits response surfaces (ordinary least squares, generalized linear models with any family including quasi-families, or linear mixed-effects models) to trait data collected under the Geometric Framework for Nutrition and computes the angular separation (theta) between the nutritional optima of two traits, with percentile bootstrap confidence intervals that propagate surface-estimation uncertainty (case or stratified resampling; paired or unpaired). Provides a design-matched coverage check that simulates validation configurations on the user's own design and at the signal-to-noise ratio of the fitted surfaces, an equivalence test for alignment of optima, local Gaussian and mean curvature, and a surface-area ratio. Includes the Drosophila lifespan and fecundity dataset of Lee et al. (2008) "Lifespan and reproduction in Drosophila: new insights from nutritional geometry" <<a href="https://doi.org/10.1073%2Fpnas.0710787105" target="_top">doi:10.1073/pnas.0710787105</a>> as a worked example. The angular approach was introduced in Morimoto (2023) "Nutrigonometry I: using right-angle triangles to quantify nutritional trade-offs in performance landscapes" <<a href="https://doi.org/10.1086%2F723599" target="_top">doi:10.1086/723599</a>>.
CRAN.R-project.org
November 8, 2026 at 12:08 AM
New on CRAN: nutrigon (0.1.0). View at https://CRAN.R-project.org/package=nutrigon