hyper parameters tweaking
play

Hyper-parameters/Tweaking Yufeng Ma, Chris Dusold Virginia Tech - PowerPoint PPT Presentation

Feb 07, 2023 •235 likes •1.29k views

Hyper-parameters/Tweaking Yufeng Ma, Chris Dusold Virginia Tech November 17, 2015 Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 1 / 40 Overview Batch Normalization 1 Internal Covariate Shift

  1. Key Points in Batch Normalization Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 13 / 40

  2. Key Points in Batch Normalization Original parameters and newly introduced γ and β will be trained. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 13 / 40

  3. Key Points in Batch Normalization Original parameters and newly introduced γ and β will be trained. When in inference, the whole population of training data is used for mean and variance statistics instead of the estimate. E ( x ) ← E B [ µ B ] m m − 1 E B [ σ 2 Var [ x ] ← B ] Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 13 / 40

  4. Key Points in Batch Normalization Original parameters and newly introduced γ and β will be trained. When in inference, the whole population of training data is used for mean and variance statistics instead of the estimate. E ( x ) ← E B [ µ B ] m m − 1 E B [ σ 2 Var [ x ] ← B ] In Convolutional layers, different locations of a feature map should be normalized in the same way. m ′ = |B| = m · pq , and γ ( k ) , β ( k ) per feature map Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 13 / 40

  5. Key Points in Batch Normalization Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 14 / 40

  6. Key Points in Batch Normalization Higher learning rates are allowed Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 14 / 40

  7. Key Points in Batch Normalization Higher learning rates are allowed BN ( Wu ) = BN (( aW ) u ) Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 14 / 40

  8. Key Points in Batch Normalization Higher learning rates are allowed BN ( Wu ) = BN (( aW ) u ) ∂ BN ( Wu ) = ∂ BN (( aW ) u ) , ∂ BN ( Wu ) = 1 a · ∂ BN (( aW ) u ) ∂ u ∂ u ∂ aW ∂ W Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 14 / 40

  9. Key Points in Batch Normalization Higher learning rates are allowed BN ( Wu ) = BN (( aW ) u ) ∂ BN ( Wu ) = ∂ BN (( aW ) u ) , ∂ BN ( Wu ) = 1 a · ∂ BN (( aW ) u ) ∂ u ∂ u ∂ aW ∂ W Batch Normalization will regularize the model with less overfitting. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 14 / 40

  10. Overview Batch Normalization 1 Internal Covariate Shift Mini-Batch Normalization Key Points in Batch Normalization Experiments and Results Importance of Initialization and Momentum 2 Overview of first-order method Momentum & Nesterov’s Accelerated Gradient(NAG) Deep Autoencoders & RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 15 / 40

  11. Activations over time Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 16 / 40

  12. Activations over time Batch Normalization helps train faster and achieve higher accuracy. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 16 / 40

  13. Activations over time Batch Normalization helps train faster and achieve higher accuracy. figure credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 16 / 40

  14. Activations over time Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 17 / 40

  15. Activations over time Batch Normalization makes input distribution more stable. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 17 / 40

  16. Activations over time Batch Normalization makes input distribution more stable. figure credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 17 / 40

  17. Accelerating Batch Normalization Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 18 / 40

  18. Accelerating Batch Normalization Networks Tricks to follow Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 18 / 40

  19. Accelerating Batch Normalization Networks Tricks to follow Increasing learning rate Remove or Reduce Dropout Reduce ℓ 2 weight regularization Accelerate the learning rate decay Remove Local Response Normalization Shuffle training examples more thoroughly Reduce the photometric distortions Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 18 / 40

  20. Network Comparisons Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 19 / 40

  21. Network Comparisons Inception, BN-Baseline, BN-x5, BN-x30, BN-x5-Sigmoid Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 19 / 40

  22. Network Comparisons Inception, BN-Baseline, BN-x5, BN-x30, BN-x5-Sigmoid Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 19 / 40

  23. Network Comparisons Inception, BN-Baseline, BN-x5, BN-x30, BN-x5-Sigmoid figure credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 19 / 40

  24. Ensemble Classification Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 20 / 40

  25. Ensemble Classification Top-5 validation error of 4.9% and test error of 4.82%, exceeds the estimated accuracy of human raters. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 20 / 40

  26. Ensemble Classification Top-5 validation error of 4.9% and test error of 4.82%, exceeds the estimated accuracy of human raters. figure credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 20 / 40

  27. Overview Batch Normalization 1 Internal Covariate Shift Mini-Batch Normalization Key Points in Batch Normalization Experiments and Results Importance of Initialization and Momentum 2 Overview of first-order method Momentum & Nesterov’s Accelerated Gradient(NAG) Deep Autoencoders & RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 21 / 40

  28. Challenges to be solved Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 22 / 40

  29. Challenges to be solved Reference paper: On the importance of initialization and momentum in deep learning Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 22 / 40

  30. Challenges to be solved Reference paper: On the importance of initialization and momentum in deep learning Difficult to use first-order method to reach performance previously only achievable by second-order method like Hessian-Free. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 22 / 40

  31. Challenges to be solved Reference paper: On the importance of initialization and momentum in deep learning Difficult to use first-order method to reach performance previously only achievable by second-order method like Hessian-Free. Well-designed random initialization Slowly increasing schedule for momentum parameter Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 22 / 40

  32. Challenges to be solved Reference paper: On the importance of initialization and momentum in deep learning Difficult to use first-order method to reach performance previously only achievable by second-order method like Hessian-Free. Well-designed random initialization Slowly increasing schedule for momentum parameter No need for sophisticated second-order methods. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 22 / 40

  33. Overview of first-order method First-order Methods Vanilla Stochastic Gradient Descent SGD + Momentum Nesterov’s Accelerated Gradient(NAG) AdaGrad Adam Rprop RMSProp AdaDelta slide credit: Ishan Misra Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 23 / 40

  34. Overview Batch Normalization 1 Internal Covariate Shift Mini-Batch Normalization Key Points in Batch Normalization Experiments and Results Importance of Initialization and Momentum 2 Overview of first-order method Momentum & Nesterov’s Accelerated Gradient(NAG) Deep Autoencoders & RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 24 / 40

  35. Several First-order Methods Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 25 / 40

  36. Several First-order Methods Notation: θ - Parameters of network, f - Objective function, ǫ - Learning rate ▽ f - Gradient of f , v - Velocity vector, µ - Momentum coefficient Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 25 / 40

  37. Several First-order Methods Notation: θ - Parameters of network, f - Objective function, ǫ - Learning rate ▽ f - Gradient of f , v - Velocity vector, µ - Momentum coefficient Vanilla SGD v t +1 = ǫ ▽ f ( θ t ) θ t +1 = θ t − v t +1 slide credit: Ishan Misra Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 25 / 40

  38. Several First-order Methods Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 26 / 40

  39. Several First-order Methods Rprop Update if ▽ f t ▽ f t − 1 > 0 v t = η + v t − 1 else if ▽ f t ▽ f t − 1 < 0 v t = η − v t − 1 else v t = v t − 1 θ t +1 = θ t − v t where 0 < η − < 1 < η + slide credit: Ishan Misra Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 26 / 40

  40. Several First-order Methods Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 27 / 40

  41. Several First-order Methods AdaGrad r t = θ 2 t + r t − 1 α v t +1 = √ r t ▽ f ( θ t ) θ t +1 = θ t − v t +1 Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 27 / 40

  42. Several First-order Methods AdaGrad r t = θ 2 t + r t − 1 α v t +1 = √ r t ▽ f ( θ t ) θ t +1 = θ t − v t +1 RMSProp = Rprop + SGD r t = (1 − γ ) θ 2 t + γ r t − 1 α v t +1 = √ r t ▽ f ( θ t ) θ t +1 = θ t − v t +1 slide credit: Ishan Misra Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 27 / 40

  43. Several First-order Methods Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 28 / 40

  44. Several First-order Methods AdaDelta v t +1 = H − 1 ▽ f , ∝ f ′ f ′′ 1 / units of θ ∝ (1 / units of θ ) 2 ∝ units of θ Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 28 / 40

  45. Several First-order Methods AdaDelta v t +1 = H − 1 ▽ f , ∝ f ′ f ′′ 1 / units of θ ∝ (1 / units of θ ) 2 ∝ units of θ Adam r t = (1 − γ 1 ) ▽ f ( θ t ) + γ 1 r t − 1 p t = (1 − γ 2 ) ▽ f ( θ t ) 2 + γ 2 p t − 1 r t r t = ˆ (1 − (1 − γ 1 ) t ) p t p t = ˆ (1 − (1 − r 2 ) t ) v t = α ˆ r t √ ˆ p t θ t +1 = θ t − v t slide credit: Ishan Misra Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 28 / 40

  46. Overview Batch Normalization 1 Internal Covariate Shift Mini-Batch Normalization Key Points in Batch Normalization Experiments and Results Importance of Initialization and Momentum 2 Overview of first-order method Momentum & Nesterov’s Accelerated Gradient(NAG) Deep Autoencoders & RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 29 / 40

  47. Momentum and NAG Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 30 / 40

  48. Momentum and NAG Notation: θ - Parameters of network, f - Objective function, ǫ - Learning rate ▽ f - Gradient of f , v - Velocity vector, µ - Momentum coefficient Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 30 / 40

  49. Momentum and NAG Notation: θ - Parameters of network, f - Objective function, ǫ - Learning rate ▽ f - Gradient of f , v - Velocity vector, µ - Momentum coefficient Classical Momentum v t +1 = µ v t − ǫ ▽ f ( θ t ) θ t +1 = θ t + v t +1 Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 30 / 40

  50. Momentum and NAG Notation: θ - Parameters of network, f - Objective function, ǫ - Learning rate ▽ f - Gradient of f , v - Velocity vector, µ - Momentum coefficient Classical Momentum v t +1 = µ v t − ǫ ▽ f ( θ t ) θ t +1 = θ t + v t +1 Nesterov’s Accelerated Gradient v t +1 = µ v t − ǫ ▽ f ( θ t + µ v t ) θ t +1 = θ t + v t +1 Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 30 / 40

  51. Relationship between CM and NAG Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 31 / 40

  52. Relationship between CM and NAG NAG uses θ t + µ v t but MISSING the yet unknown correction. Thus when the addition of µ v t results in an immediate undesirable increase in the objective f , Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 31 / 40

  53. Relationship between CM and NAG NAG uses θ t + µ v t but MISSING the yet unknown correction. Thus when the addition of µ v t results in an immediate undesirable increase in the objective f , Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 31 / 40

  54. Relationship between CM and NAG NAG uses θ t + µ v t but MISSING the yet unknown correction. Thus when the addition of µ v t results in an immediate undesirable increase in the objective f , figure credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 31 / 40

  55. Relationship between CM and NAG Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 32 / 40

  56. Relationship between CM and NAG Apply CM and NAG to a positive definite quadratic objective q ( x ) = x T Ax / 2 + b T x . Difference in effective momentum coefficient Classical Momentum: µ NAG: µ (1 − λǫ ), where λ is the eigenvalue of A . Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 32 / 40

  57. Relationship between CM and NAG Apply CM and NAG to a positive definite quadratic objective q ( x ) = x T Ax / 2 + b T x . Difference in effective momentum coefficient Classical Momentum: µ NAG: µ (1 − λǫ ), where λ is the eigenvalue of A . ǫ small, CM and NAG are equivalent ǫ large, NAG gives smaller µ (1 − λ i ǫ ) to stop oscillations. Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 32 / 40

  58. Overview Batch Normalization 1 Internal Covariate Shift Mini-Batch Normalization Key Points in Batch Normalization Experiments and Results Importance of Initialization and Momentum 2 Overview of first-order method Momentum & Nesterov’s Accelerated Gradient(NAG) Deep Autoencoders & RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 33 / 40

  59. Deep Autoencoders Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 34 / 40

  60. Deep Autoencoders Structure of Deep Autoencoder Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 34 / 40

  61. Deep Autoencoders Structure of Deep Autoencoder figure credit: http://deeplearning4j.org/deepautoencoder.html Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 34 / 40

  62. Deep Autoencoders Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 35 / 40

  63. Deep Autoencoders Sparse Initialization -each random unit connected to 15 randomly chosen units in the previous layer, drawn from a unit Gaussian Schedule for Momentum Coefficient µ t = min (1 − 2 − 1 − log 2 ( ⌊ t / 250 ⌋ +1) , µ max ) Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 35 / 40

  64. Deep Autoencoders Sparse Initialization -each random unit connected to 15 randomly chosen units in the previous layer, drawn from a unit Gaussian Schedule for Momentum Coefficient µ t = min (1 − 2 − 1 − log 2 ( ⌊ t / 250 ⌋ +1) , µ max ) µ t = 1 − 3 / ( t + 5), not strongly convex - Nesterov(1983) constant µ t , strongly convex - Nesterov(2003) Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 35 / 40

  65. Deep Autoencoders Sparse Initialization -each random unit connected to 15 randomly chosen units in the previous layer, drawn from a unit Gaussian Schedule for Momentum Coefficient µ t = min (1 − 2 − 1 − log 2 ( ⌊ t / 250 ⌋ +1) , µ max ) µ t = 1 − 3 / ( t + 5), not strongly convex - Nesterov(1983) constant µ t , strongly convex - Nesterov(2003) table credit: reference paper Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 35 / 40

  66. RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 36 / 40

  67. RNN - Echo-State Networks Echo-State Networks (a family RNNs) Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 36 / 40

  68. RNN - Echo-State Networks Echo-State Networks (a family RNNs) figure credit: Mantas Lukoevicius Hidden-to-output connections learned from data Recurrent connections fixed to a random draw from a specific distribution Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 36 / 40

  69. RNN - Echo-State Networks Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 37 / 40

  70. RNN - Echo-State Networks ESN-based Initialization Spectral Radius of Hidden-to-hidden matrix around 1.1 Initial scale of Input-to-hidden connections plays an important role (Gaussian draw with a standard deviation of 0.001 achieves good balance, but is Task Dependent) Yufeng Ma, Chris Dusold (Virginia Tech) Hyper-parameters/Tweaking November 17, 2015 37 / 40

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Hyper: Make VM Runs Like Container Xu Wang &lt;xu@hyper.sh&gt; Hyper HQ Agenda Lesson

Hyper: Make VM Runs Like Container Xu Wang &lt;xu@hyper.sh&gt; Hyper HQ Agenda Lesson

Hyper: Make VM Runs Like Container Xu Wang &lt;xu@hyper.sh&gt; Hyper HQ Agenda Lesson learned from docker Hyper: App-centric VM Hyper and Xen Next step Docker Seems to Beat VM Everywhere Docker announced in 2013 and release

531 views • 13 slides
Tweaking structures: working on the fiddly bits Kevin Karplus karplus@soe.ucsc.edu Biomolecular

Tweaking structures: working on the fiddly bits Kevin Karplus karplus@soe.ucsc.edu Biomolecular

Tweaking structures: working on the fiddly bits Kevin Karplus karplus@soe.ucsc.edu Biomolecular Engineering Department University of California, Santa Cruz Tweaking structures p.1/12 Outline of Talk Cost functions Clash detection

334 views • 12 slides
From HyPer to Hyper Integrating an academic DBMS into a leading analytics and business

From HyPer to Hyper Integrating an academic DBMS into a leading analytics and business

From HyPer to Hyper Integrating an academic DBMS into a leading analytics and business intelligence platform Tobias Muehlbauer, tmuehlbauer@tableau.com Jan Finis, jfinis@tableau.com The Story of Hyper 2020: Hyper as a general-purpose Database

747 views • 37 slides
RFC6962-bis update Some tweaking still needed. Last call should wait for our

RFC6962-bis update Some tweaking still needed. Last call should wait for our

RFC6962-bis update Some tweaking still needed. Last call should wait for our implementation. A member of the Certificate Transparency team at Google is working on it. We think all major issues resolved.

258 views • 10 slides
Tra racking Hyper Bo cking Hyper Boosted sted Top Q p Qua uarks @ 100 T rks @ 100 TeV eV

Tra racking Hyper Bo cking Hyper Boosted sted Top Q p Qua uarks @ 100 T rks @ 100 TeV eV

Tra racking Hyper Bo cking Hyper Boosted sted Top Q p Qua uarks @ 100 T rks @ 100 TeV eV Michel Mi chele Sel e Selva vaggi ggi (CP3) (CP3) An Andr drew w Lar Larko koski ski (MIT), Fabi abio Mal altoni (CP CP3) Flavo vor

623 views • 31 slides
Hyper-Resolution AUTOMATED REASONING Hyper-resolution is the strategy employed (electron) in the

Hyper-Resolution AUTOMATED REASONING Hyper-resolution is the strategy employed (electron) in the

8ai Hyper-Resolution AUTOMATED REASONING Hyper-resolution is the strategy employed (electron) in the widely used Otter family of provers. SLIDES 8: {Px,Qx,Rx} {Qa,C} Hyper-resolution generalises ``bottom- (nucleus) up reasoning

343 views • 7 slides
Outreach for Hyper-Kamiokande Jost Migenda (they/them) for the Hyper-Kamiokande

Outreach for Hyper-Kamiokande Jost Migenda (they/them) for the Hyper-Kamiokande

Outreach for Hyper-Kamiokande Jost Migenda (they/them) for the Hyper-Kamiokande Proto-Collaboration @JostMigenda TAUP 2019 2019-09-11 Disclaimer This talk is not exhaustive! Personal, subjective selection Thanks to the whole

576 views • 20 slides
Status of the Hyper- Kamiokande Experiment Erin OSullivan, on behalf of the Hyper-Kamiokande

Status of the Hyper- Kamiokande Experiment Erin OSullivan, on behalf of the Hyper-Kamiokande

Status of the Hyper- Kamiokande Experiment Erin OSullivan, on behalf of the Hyper-Kamiokande proto-collaboration Stockholm University NuFACT 2017 Hyper-Kamiokande collaboration Proto-collaboration formed in 2015 ~ 300 members from 14

789 views • 31 slides
Tweaking Code-Based Cryptography for Embedded Systems DIMACS Workshop on The Mathematics of

Tweaking Code-Based Cryptography for Embedded Systems DIMACS Workshop on The Mathematics of

Tweaking Code-Based Cryptography for Embedded Systems DIMACS Workshop on The Mathematics of Post-Quantum Cryptography Tim Gneysu, Ingo von Maurich 1/12/2015 Horst Grtz Institute for IT-Security, Ruhr-Universitt Bochum, Germany Motivation

920 views • 70 slides
Quantum Lattice Enumeration and Tweaking Discrete Pruning Yoshinori Aono Phong Q.

Quantum Lattice Enumeration and Tweaking Discrete Pruning Yoshinori Aono Phong Q.

Quantum Lattice Enumeration and Tweaking Discrete Pruning Yoshinori Aono Phong Q. Nguyen Yixin Shen ASIACRYPT 2018 Context NIST standardization of post-quantum cryptography: Need to convince security

232 views • 22 slides
Hyper-Resolution AUTOMATED REASONING Hyper-resolution generalises ``bottom- (electron) up

Hyper-Resolution AUTOMATED REASONING Hyper-resolution generalises ``bottom- (electron) up

8ai Hyper-Resolution AUTOMATED REASONING Hyper-resolution generalises ``bottom- (electron) up reasoning and combines several Px Qx Rx Qa C SLIDES 8: resolution steps into one big step. (nucleus) Hyper-resolution is the

372 views • 8 slides
Indirect WIMP detection with neutrinos in Hyper-K Yusuke Koshio for Hyper-K astrophysics working

Indirect WIMP detection with neutrinos in Hyper-K Yusuke Koshio for Hyper-K astrophysics working

Indirect WIMP detection with neutrinos in Hyper-K Yusuke Koshio for Hyper-K astrophysics working group Kamioka observatory, ICRR, Univ. of Tokyo Cosmic Frontier Workshop SLAC - March 7, 2013 Neutrinos generated by WIMP annihilation Trap

542 views • 20 slides
Neutrino parameters with large experiments (CP violation, mass hierarchy,...) Present Status

Neutrino parameters with large experiments (CP violation, mass hierarchy,...) Present Status

Mauro Mezzetto, Istituto Nazionale Fisica Nucleare, Padova Neutrino parameters with large experiments (CP violation, mass hierarchy,...) Present Status Different approaches for mass hierarchy Measuring CP: Dune and Hyper-Kamiokande

426 views • 29 slides
Gibbs Sampling for LDA Lei Tang Department of CSE Arizona State University January 7, 2008 1 /

Gibbs Sampling for LDA Lei Tang Department of CSE Arizona State University January 7, 2008 1 /

Gibbs Sampling for LDA Lei Tang Department of CSE Arizona State University January 7, 2008 1 / 10 Graphical Representation , are fixed hyper-parameters. We need to estimate parameters for each document and for each topic. Z are

483 views • 11 slides
Hyper-K David Hadley, University of Warwick Outline Hyper-K Detector Long baseline neutrino

Hyper-K David Hadley, University of Warwick Outline Hyper-K Detector Long baseline neutrino

Hyper-K David Hadley, University of Warwick Outline Hyper-K Detector Long baseline neutrino oscillation status and prospects Systematic uncertainty challenges and solutions 2 Kamiokande Detectors Kamiokande 680 tonne fiducial mass (1983)

1.05k views • 77 slides
Tweaking Even-Mansour Ciphers Benot Cogliati 1 Rodolphe Lampe 1 Yannick Seurin 2 1 Versailles

Tweaking Even-Mansour Ciphers Benot Cogliati 1 Rodolphe Lampe 1 Yannick Seurin 2 1 Versailles

Tweakable Block Ciphers Our Contribution Proof Overview Longer Cascades Conclusion Tweaking Even-Mansour Ciphers Benot Cogliati 1 Rodolphe Lampe 1 Yannick Seurin 2 1 Versailles University, France 2 ANSSI, France August 17, 2015 CRYPTO

784 views • 77 slides
Developing a RESTful Web application for Liberty in CICS Introduction Course introduction What

Developing a RESTful Web application for Liberty in CICS Introduction Course introduction What

Developing a RESTful Web application for Liberty in CICS Introduction Course introduction What youll see in this course Eclipse development environment RESTful web application CICS region Liberty JVM server HTTP Call JCICS RESTful

1.36k views • 118 slides
SWEN 256 Software Process &amp; Project Management Not everything that can be

SWEN 256 Software Process &amp; Project Management Not everything that can be

SWEN 256 Software Process &amp; Project Management Not everything that can be counted counts, and not everything that counts can be counted. - Albert Einstein Software measurement is concerned with deriving a quantit

547 views • 30 slides
Welcome! A Case for Response Time Focused Query Processing Olaf Hartjg

Welcome! A Case for Response Time Focused Query Processing Olaf Hartjg

Welcome! A Case for Response Time Focused Query Processing Olaf Hartjg @olafiartjg 1 Informatjon in Dynamic Web Pages A Case for Response Time Focused Query Processing Olaf Hartjg

1.16k views • 84 slides
CSCI 3 3 4 2 I nternet Program m ing MW , 3 :0 5 pm 4 :2 0 pm ACSB 2 .1 1 3 Emmett Tomai

CSCI 3 3 4 2 I nternet Program m ing MW , 3 :0 5 pm 4 :2 0 pm ACSB 2 .1 1 3 Emmett Tomai

CSCI 3 3 4 2 I nternet Program m ing MW , 3 :0 5 pm 4 :2 0 pm ACSB 2 .1 1 3 Emmett Tomai Office: ENGR 3.2100 Phone: 665-7229 Email: emmett.tomai@utrgv.edu Course Information http://faculty.utrgv.edu/emmett.tomai/courses/3342/ All

327 views • 15 slides
Neural Autoregressive Distribution Estimation Instructor: John Thickstun Discussion Board:

Neural Autoregressive Distribution Estimation Instructor: John Thickstun Discussion Board:

Neural Autoregressive Distribution Estimation Instructor: John Thickstun Discussion Board: Available on Ed! Zoom Link: Available on Canvas Instructor Contact: thickstn@cs.washington.edu Course Webpage:

842 views • 18 slides
Boo ook Acq cquis isitio ion in n the the (ne (new) era er a of of dig digit ital l

Boo ook Acq cquis isitio ion in n the the (ne (new) era er a of of dig digit ital l

Boo ook Acq cquis isitio ion in n the the (ne (new) era er a of of dig digit ital l ed educatio ion (Edit/crop photo to align within this space) 3 June 2020 Zoe Loveland, Sr Director Regional Marketing EMEAI Cherry Otto, Market

520 views • 35 slides
Second Order Adjoints with the NAG Fortran 95 Compiler Uwe Naumann, Michael Maier RWTH Aachen,

Second Order Adjoints with the NAG Fortran 95 Compiler Uwe Naumann, Michael Maier RWTH Aachen,

Second Order Adjoints with the NAG Fortran 95 Compiler Uwe Naumann, Michael Maier RWTH Aachen, Germany Bruce Christianson, Dmitrij Gendler, Jan Riehme University of Hertfordshire, UK riehme@stce.rwth-aachen.de May 22, 2007 Outline COMPAD-II

1.29k views • 61 slides
On the Verification of Synthesized Kalman Filters Ruben Gamboa, John Cowles, Jeff Van Baalen

On the Verification of Synthesized Kalman Filters Ruben Gamboa, John Cowles, Jeff Van Baalen

1 On the Verification of Synthesized Kalman Filters Ruben Gamboa, John Cowles, Jeff Van Baalen University of Wyoming ACL2 Workshop 2003 Supported by NASA grant NAG 2-1570 2 The General Challenge Consider the automatic generation of

815 views • 24 slides

More recommend