Support Vector Machines: A More Certain Estimate of Uncertainty in Regression Tasks
Pritam Anand
Action editor: Matias Valdenegro-Toro
https://openreview.net/forum?id=ZdzIDoGeOC
#svms #svm #prediction
Pritam Anand
Action editor: Matias Valdenegro-Toro
https://openreview.net/forum?id=ZdzIDoGeOC
#svms #svm #prediction
November 8, 2026 at 9:21 PM
Support Vector Machines: A More Certain Estimate of Uncertainty in Regression Tasks
Pritam Anand
Action editor: Matias Valdenegro-Toro
https://openreview.net/forum?id=ZdzIDoGeOC
#svms #svm #prediction
Pritam Anand
Action editor: Matias Valdenegro-Toro
https://openreview.net/forum?id=ZdzIDoGeOC
#svms #svm #prediction
Eine systematische Analyse zeigt, dass die Readout-Schicht der primäre Engpass für die Genauigkeit in variationellen Quantenklassifizierern zur Eindringlingserkennung ist. Quantenschaltungen entsprechen der Leistung klassischer SVMs; der Quantenvorteil bleibt für diese Aufgabe...
Variationelle Quantenklassifizierer: Wo die Genauigkeit bei der Cyber-Angriffserkennung verloren geht
arxiv.org
November 3, 2026 at 4:58 AM
Eine systematische Analyse zeigt, dass die Readout-Schicht der primäre Engpass für die Genauigkeit in variationellen Quantenklassifizierern zur Eindringlingserkennung ist. Quantenschaltungen entsprechen der Leistung klassischer SVMs; der Quantenvorteil bleibt für diese Aufgabe...
The model is calibrated. Any rewrite is destructive.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
October 31, 2026 at 4:31 PM
The model is calibrated. Any rewrite is destructive.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
I've been working on a personal data science project. In this case, the goal is to compare View of the Hebrews against the Book of Mormon and another writing known to be produced by Joseph Smith, Jr. and company. My plan is to use KNN, and to use SVMs to see if BofM was plagiarized.
October 28, 2026 at 2:20 AM
I've been working on a personal data science project. In this case, the goal is to compare View of the Hebrews against the Book of Mormon and another writing known to be produced by Joseph Smith, Jr. and company. My plan is to use KNN, and to use SVMs to see if BofM was plagiarized.
Die Liste ist ein wenig willkürlich. Mir fallen etwa 10 bedeutende Algorithmen in der Periode ein. Zum Beispiel "Support Vector Machines (SVMs)".
October 23, 2026 at 2:00 AM
Die Liste ist ein wenig willkürlich. Mir fallen etwa 10 bedeutende Algorithmen in der Periode ein. Zum Beispiel "Support Vector Machines (SVMs)".
New on CRAN: l1penSVMs (0.1.0). View at https://CRAN.R-project.org/package=l1penSVMs
l1penSVMs: Regularization Paths of L1-Penalized SVMs
Computes regularization paths for L1-penalized support vector machine (SVM) models, including the standard SVM, receiver operating characteristic SVM (ROC-SVM), and fraud detection SVM (FD-SVM). The package implements efficient path algorithms to obtain solutions over a wide range of regularization parameter values. It also provides cross-validation for selecting an optimal regularization parameter, along with functions for prediction, visualization, and summarization of the fitted models. For more details, see Zhu et al. (2003) <<a href="https://proceedings.neurips.cc/paper_files/paper/2003/file/49d4b2faeb4b7b9e745775793141e2b2-Paper.pdf" target="_top">https://proceedings.neurips.cc/paper_files/paper/2003/file/49d4b2faeb4b7b9e745775793141e2b2-Paper.pdf</a>>, Kim et al. (2021) <<a href="https://doi.org/10.1002%2Fsta4.400" target="_top">doi:10.1002/sta4.400</a>>, Park et al. (2023) <<a href="https://doi.org/10.1007%2Fs42952-023-00207-6" target="_top">doi:10.1007/s42952-023-00207-6</a>>.
CRAN.R-project.org
October 21, 2026 at 8:27 PM
New on CRAN: l1penSVMs (0.1.0). View at https://CRAN.R-project.org/package=l1penSVMs
Todays mobile suit: SVMS-01 Union Flag
October 17, 2026 at 12:15 PM
Todays mobile suit: SVMS-01 Union Flag
I LOOOOOOOOOOOOOOOOOOOVE THE SVMS-01 UNION FLAG
Todays mobile suit: SVMS-01 Union Flag
October 17, 2026 at 12:16 PM
I LOOOOOOOOOOOOOOOOOOOVE THE SVMS-01 UNION FLAG
Hitanshi Arora, Mohit Swarnkar, Aswani Kumar Cherukuri: Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints https://arxiv.org/abs/2610.17027 https://arxiv.org/pdf/2610.17027 https://arxiv.org/html/2610.17027
October 15, 2026 at 6:42 AM
Hitanshi Arora, Mohit Swarnkar, Aswani Kumar Cherukuri: Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints https://arxiv.org/abs/2610.17027 https://arxiv.org/pdf/2610.17027 https://arxiv.org/html/2610.17027
Hitanshi Arora, Mohit Swarnkar, Aswani Kumar Cherukuri: Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints https://arxiv.org/abs/2610.17027 https://arxiv.org/pdf/2610.17027 https://arxiv.org/html/2610.17027
October 14, 2026 at 6:42 AM
Hitanshi Arora, Mohit Swarnkar, Aswani Kumar Cherukuri: Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints https://arxiv.org/abs/2610.17027 https://arxiv.org/pdf/2610.17027 https://arxiv.org/html/2610.17027
Today I have been at UoN for 12 months, 5 months of which I have been "at risk of redundancy". Despite the School I am in not making any redundancies, they actually need to recruit to lower the staff-student ratio for accreditation, I am one of 7 SVMS academics in a pool with people form another...
October 6, 2026 at 10:15 AM
Today I have been at UoN for 12 months, 5 months of which I have been "at risk of redundancy". Despite the School I am in not making any redundancies, they actually need to recruit to lower the staff-student ratio for accreditation, I am one of 7 SVMS academics in a pool with people form another...
When I was in school (the mid aughts) “AI” was neural nets and SVMs and decision trees, and the older symbolic methods, which were played out, and ML was (a few years later) the new statistical methods
October 6, 2026 at 1:17 AM
When I was in school (the mid aughts) “AI” was neural nets and SVMs and decision trees, and the older symbolic methods, which were played out, and ML was (a few years later) the new statistical methods
Hell yeah, SVMs ahoy
October 4, 2026 at 7:26 PM
Hell yeah, SVMs ahoy
not sure if there is a good unifying frame--in the 80s and 90s "cognitive architectures" (Soar, ACT-R, etc) and multi-agent systems were popular and you can kinda cram that into a utility maximizer framework. by 2005, it was all about SVMs and gradient boosted trees and stuff. by 2015, it's all DL
October 4, 2026 at 9:47 PM
not sure if there is a good unifying frame--in the 80s and 90s "cognitive architectures" (Soar, ACT-R, etc) and multi-agent systems were popular and you can kinda cram that into a utility maximizer framework. by 2005, it was all about SVMs and gradient boosted trees and stuff. by 2015, it's all DL
People are going to learn about the mathematical equivalence of SVMs and Rosenberg perceptrons and get super mad at some point, one assumes
October 2, 2026 at 12:34 AM
People are going to learn about the mathematical equivalence of SVMs and Rosenberg perceptrons and get super mad at some point, one assumes
Quantum-Kernel-SVMs zeigen aufgabenabhängige Leistung für die Klassifizierung von Netzwerksicherheit—wettbewerbsfähig bei einigen Datensätzen, aber nicht universell überlegen. Quantum-Oracle-Sketching deutet auf langfristige Vorteile der Speichereffizienz gegenüber explizitem ...
Quantum-Kernel und Oracle-Sketching für die Klassifizierung von Netzwerksicherheit: Kurzfristige und langfristige Analyse
arxiv.org
September 30, 2026 at 4:05 AM
Quantum-Kernel-SVMs zeigen aufgabenabhängige Leistung für die Klassifizierung von Netzwerksicherheit—wettbewerbsfähig bei einigen Datensätzen, aber nicht universell überlegen. Quantum-Oracle-Sketching deutet auf langfristige Vorteile der Speichereffizienz gegenüber explizitem ...
Quantum kernel SVMs show task-dependent performance for network-security classification—competitive on some datasets but not universally superior. Quantum oracle sketching suggests long-term memory efficiency advantages over explicit sparse storage for large-scale data access.
Quantum Kernels and Oracle Sketching for Network Security Classification: Near-Term and Long-Term Analysis
arxiv.org
September 30, 2026 at 4:03 AM
Quantum kernel SVMs show task-dependent performance for network-security classification—competitive on some datasets but not universally superior. Quantum oracle sketching suggests long-term memory efficiency advantages over explicit sparse storage for large-scale data access.
क्वांटम कर्नेल SVMs नेटवर्क-सुरक्षा वर्गीकरण के लिए कार्य-आश्रित प्रदर्शन दिखाते हैं—कुछ डेटासेट पर प्रतिस्पर्धी लेकिन सार्वभौमिक रूप से श्रेष्ठ नहीं। क्वांटम ओरेकल स्केचिंग बड़े पैमाने पर डेटा एक्सेस के लिए स्पष्ट विरल भंडारण पर दीर्घकालीन मेमोरी दक्षता लाभ का सुझाव देता है।
क्वांटम कर्नेल और नेटवर्क सुरक्षा वर्गीकरण के लिए ओरेकल स्केचिंग: निकट-अवधि और दीर्घकालीन विश्लेषण
arxiv.org
September 30, 2026 at 4:06 AM
क्वांटम कर्नेल SVMs नेटवर्क-सुरक्षा वर्गीकरण के लिए कार्य-आश्रित प्रदर्शन दिखाते हैं—कुछ डेटासेट पर प्रतिस्पर्धी लेकिन सार्वभौमिक रूप से श्रेष्ठ नहीं। क्वांटम ओरेकल स्केचिंग बड़े पैमाने पर डेटा एक्सेस के लिए स्पष्ट विरल भंडारण पर दीर्घकालीन मेमोरी दक्षता लाभ का सुझाव देता है।
SVMs de núcleos quânticos mostram desempenho dependente da tarefa para classificação de segurança de rede—competitivos em alguns conjuntos de dados, mas não universalmente superiores. O esboço de oráculo quântico sugere vantagens de eficiência de memória de longo prazo sobre a...
Núcleos Quânticos e Esboço de Oráculo para Classificação de Segurança de Rede: Análise de Curto Prazo e Longo Prazo
arxiv.org
September 30, 2026 at 4:06 AM
SVMs de núcleos quânticos mostram desempenho dependente da tarefa para classificação de segurança de rede—competitivos em alguns conjuntos de dados, mas não universalmente superiores. O esboço de oráculo quântico sugere vantagens de eficiência de memória de longo prazo sobre a...
Las SVMs de núcleos cuánticos muestran un rendimiento dependiente de la tarea para la clasificación de seguridad de redes, competitivo en algunos conjuntos de datos pero no universalmente superior. El bosquejo de oráculos cuánticos sugiere ventajas de eficiencia de memoria a l...
Núcleos Cuánticos y Bosquejo de Oráculos para la Clasificación de Seguridad de Redes: Análisis a Corto y Largo Plazo
arxiv.org
September 30, 2026 at 4:05 AM
Las SVMs de núcleos cuánticos muestran un rendimiento dependiente de la tarea para la clasificación de seguridad de redes, competitivo en algunos conjuntos de datos pero no universalmente superior. El bosquejo de oráculos cuánticos sugiere ventajas de eficiencia de memoria a l...
ควอนตัมเคอร์เนล SVMs แสดงประสิทธิภาพที่ขึ้นอยู่กับงานสำหรับการจำแนกความปลอดภัยเครือข่าย—แข่งขันกันได้ในบางชุดข้อมูล แต่ไม่ได้เหนือกว่าโดยทั่วไป ควอนตัมออราเคิลสเก็ตชิ่งแสดงให้เห็นประโยชน์ของประสิทธิภาพหน่วยความจำระยะยาวเหนือการจัดเก็บแบบกระจายอย่างชัดแจ้งสำหรับการเข้าถึงข้อมูล...
ควอนตัมเคอร์เนลและออราเคิลสเก็ตชิ่งสำหรับการจำแนกความปลอดภัยเครือข่าย: การวิเคราะห์ระยะสั้นและระยะยาว
arxiv.org
September 30, 2026 at 4:07 AM
ควอนตัมเคอร์เนล SVMs แสดงประสิทธิภาพที่ขึ้นอยู่กับงานสำหรับการจำแนกความปลอดภัยเครือข่าย—แข่งขันกันได้ในบางชุดข้อมูล แต่ไม่ได้เหนือกว่าโดยทั่วไป ควอนตัมออราเคิลสเก็ตชิ่งแสดงให้เห็นประโยชน์ของประสิทธิภาพหน่วยความจำระยะยาวเหนือการจัดเก็บแบบกระจายอย่างชัดแจ้งสำหรับการเข้าถึงข้อมูล...
The model is calibrated. Any rewrite is destructive.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
September 26, 2026 at 4:30 PM
The model is calibrated. Any rewrite is destructive.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
If |Z| > 3.0, use Venn-Abers — distribution-free, mathematically guaranteed. The 1.96–3.0 band is where you actually test both.
The standard advice comes from an era of SVMs and shallow nets.
SVMs are mathematically provable to be the best way to discriminate two classes of things. It was built in to programs like Apple's Mail to distinguish your spam from not-spam by you clicking in the "it's spam" button. At this point, machine learning seemed to be a dead end.
3/8
3/8
September 18, 2026 at 2:33 PM
SVMs are mathematically provable to be the best way to discriminate two classes of things. It was built in to programs like Apple's Mail to distinguish your spam from not-spam by you clicking in the "it's spam" button. At this point, machine learning seemed to be a dead end.
3/8
3/8
Cost, speed, and correctness. There’s a reason we used SVMs for a lot of things for decade or so: a lot of the time, all you really want is a yes/no answer (and the fancier ones could give you a confidence value as well).
September 18, 2026 at 10:12 AM
Cost, speed, and correctness. There’s a reason we used SVMs for a lot of things for decade or so: a lot of the time, all you really want is a yes/no answer (and the fancier ones could give you a confidence value as well).
Which is fine, I like SVMs for the most part, and I’ve long been frustrated by broad insistence on chat output where that’s not a thing that makes sense
September 18, 2026 at 2:44 AM
Which is fine, I like SVMs for the most part, and I’ve long been frustrated by broad insistence on chat output where that’s not a thing that makes sense