[DL輪読会] Spectral Norm Regularization for Improving the Generalizability of Deep Learning/Spectral Normalization for GANs

Deep Learning JP
Deep Learning JPDeep Learning JP
DEEP LEARNING JP
[DL Papers]
“Spectral Norm Regularization for Improving the
Generalizability of Deep Learning” and “Spectral
Normalization for Generative Adversarial Networks
Reiji Hatsugai
http://deeplearning.jp/
書誌情報
• Spectral Norm Regularization
– arxiv 5/31 (NIPS2017?)
– NII吉田さん、PFN宮戸さん
• Spectral Normalization for GANs
– ICML2017 Workshop on Implicit Models (https://drive.google.com/file/d/0B8HZ50DPgR3eSVV6YlF3XzQxSjQ/view?usp=sharing)
– PFN宮戸さん et al
2
モチベーション
• ニューラルネットワークとは
– 神経回路網
– 脳
– 人間より高精度の認識
– Atari、碁で人間をoverperfrom
• でも結局行列演算
– 行列の特性を考慮に入れて学習アルゴリズムを作りたい
– 行列をオペレータと見たときに自然に定義されるノルムである、
スペクトルノルムに着目している
– 汎化性能、頑健性、学習安定性の向上に効きそう
3
線形代数の復習
• いくつかの線形代数の知識を復習してから話します
– 特異値分解
– スペクトルノルム
4
特異値分解
• 特異値分解とは
– 行列分解の一つ
– 対象となる行列Wを直交行列U, Vと対角行列Σの積で分解
• UとVの列 (行) ベクトルは正規直交基底をなす
– 対角行列の成分は特異値と呼ばれる
5
𝑈" 𝑈 = 𝐼, 𝑉𝑉" = 𝐼 Σ = diag(𝜎., … , 𝜎0)
𝑊 = 𝑈ΣV4
特異値、nはWのランク、𝜎. ≥ 𝜎6 ⋯ ≥ 𝜎0
特異値分解の性質
• 特異行列 (ランク1の行列) の和に分解することができる
• 特異値は𝑊"
𝑊の固有値の平方根と一致する
• Wが正方行列だった場合固有値分解と同等
6
特異行列の和に分解
• Uの列ベクトルを𝑢., 𝑢6, … , 𝑢0
• Vの列ベクトルを𝑣., 𝑣6, … , 𝑣0とする
7
𝑊 = 𝑈ΣV4
= 𝑢. ⋯ 𝑢0
𝜎. ⋯ 0
⋮ ⋱ ⋮
0 ⋯ 𝜎0
𝑣.
"
⋮
𝑣0
"
= = 𝜎> 𝑢> 𝑣>
"
0
>
ランク1
𝑊" 𝑢> = 𝜎> 𝑣>
𝑊𝑣> = 𝜎> 𝑢>
固有値っぽい性質
特異値分解の幾何学的意味
• 𝑊 ∈ 𝑅0×B
としてVの列ベクトルがなす空間からUの列ベクトルがなす
空間への写像としてWを考えると、
• 𝑣>が対応する𝑢>に写り、その長さを𝜎>倍する
8
𝑣.
𝑣6
𝑣C 𝑊 𝑢.
𝑢6
𝑢C
𝜎C
𝜎6
𝜎.
スペクトルノルム
• 行列がある空間のベクトルに対して作用したときの
行き先でのノルムと前の空間でのノルムの比の上限
• 行列Wの最大特異値と一致!
9
𝑥 ⟼ 𝑊𝑥 sup
I
𝑊𝑥
𝑥
= sup
I J.
𝑊𝑥
𝑊𝑥 = 𝑥" 𝑊" 𝑊𝑥
= 𝑥" 𝑉ΣU4UΣV4x
= 𝑥" 𝑉ΣV4x 𝑥 = ∑ 𝛼> 𝑣>
0
> と置く
= = 𝛼>
6
𝜎>
6
0
>
𝛼>
6
が正で∑ 𝛼>
6
= 1なので𝛼. = 1,	それ以外0で
最大値𝜎.を取る
Spectral Norm Regularization
• ついに論文に入ります
• 論文要旨
– NNの各層のSNに正則化を加えることで入力データのperturbationに対して
ロバストなNNを学習し汎化性能を向上させた
10
論文モチベーション
• 汎化性能に対する一つの仮説
– 入力データの微小なnoise (perturbation) に対してモデルの出力がsensitiveになる
と学習器の性能が悪化する
• 入力データの微小な変化による出力値の変化を評価したい
• NN全体を一つの作用素と見たときの作用素ノルム (スペクトルノルム)
11
メインアイデア
• 多層NNは行列演算と非線形関数のカスケード接続
• 微小なnoiseが加えられた入力を評価する
• noiseが微小だとすると、各層のelementwiseな非線形変換は
線形変換とみなすことができ、単純なAffine変換の式で書ける
• スペクトルノルムがupper bound!
12
ロス関数と勾配計算
• 通常のロス関数にスペクトルノルムの正則項を加えたロス関数
• 勾配法で学習するため勾配の計算が必要
13
𝜕𝜎 𝑊 6
𝜕𝑊
= 2𝜎 𝑊
𝜕𝜎 𝑊
𝜕𝑊
= 2𝜎 𝑊 𝑢. 𝑣.
"
(ここよくわかっていません。妄想)
特異値分解の式から最大特異値の勾配方向は𝑢. 𝑣.
"
𝑊 = 𝜎. 𝑢. 𝑣.
"
+ 𝜎6 𝑢6 𝑣6
"
+ ⋯ + 𝜎0 𝑢0 𝑣0
"
−𝜆𝑢. 𝑣.
"
をWに加えるとその分𝜎.が下がる
スペクトルノルムの計算
• スペクトルノルムを計算するたびに特異値分解をすると計算量が増大
• power iteration法を使って高速に計算 (実際には各iterでuを保持しておき、反復は一回のみ)
14
𝑥W = = 𝛼> 𝑢>
𝑥. = 𝑊𝑦W = = 𝛼> 𝜎>
6
𝑢>
𝑦W = 𝑊" 𝑥W = = 𝛼> 𝜎> 𝑣>
𝑥0 = = 𝛼> 𝜎>
60
𝑢> , 𝑥0Y. = = 𝛼> 𝜎>
60Y6
𝑢>
⋮
𝑥0
𝜎.
60 = = 𝛼>
𝜎>
𝜎.
60
𝑢> ≈ 𝑢. ,
𝑥0Y.
𝜎.
60Y6 = = 𝛼>
𝜎>
𝜎.
60Y6
𝑢> ≈ 𝑢.
𝑥0
𝑥0Y.
= 𝜎., 𝑢. =
𝑥0
𝑥0
, 𝑣. =
𝑦0
𝑦0
結果1
15
結果2
16
結論
• spectral normに対する正則化によって汎化性能が向上したことが確認
できた
• MAP推定においてパラメタの事前分布に正規分布を仮定するとL2正
則化がでてくる。これと同じような構造がSpectral Normにもあるだと
うか?どんな事前分布だろうか?
17
Spectral Normalization for GANs
• 論文概要
– GANsの学習を安定させるにはLipschitz連続であることが重要であると最近の
研究でわかりつつある。Spectral Normalizationを行うことでDiscriminatorの
Lipschitz定数をコントロールし安定化を向上させた
18
Lipschitz連続とは
• ある作用素fで任意の2点を変換し、変換後の距離と変換前の距離の
比を抑える最小の定数をKとするとfはK-Lipschitz
• 今回はなぜGANsにおいてLipschitz性が重要であるかは割愛します。
WGAN, LSGAN, b-GAN等を見てください
19
fがK-Lipschitz⟺ ∀𝑥, 𝑥′
^ I Y^(I_)
IYI`
≤ 𝐾
LipschitzとSpectal Norm
• 非線形関数のLipschitz定数が1とするとNN全体のLipschitz定数は各層
のLipschitz定数の積で抑えられる
• 各層のLipschitz定数の積は各層のスペクトルノルムの積で表せる
20
Spectral Normalization
• Spectral Normで重みを割ってできた重みを使ってFeedForwardする
• これによってNN全体のLipschitz定数を1で抑えることができる
21
結果
22
Inception Score
頑健性
23
結論
• Spectral Normalizationによって安定性が向上した
• Inception Scoreもweight normalizationに比べて良いか同等のスコアを出
した
24
感想
• スペクトルノルムは今後画像以外の分野でも使えそう
• 学習安定性、Adversarial example、汎化性能等の最近hotな分野とちか
い感じがする
• 強化学習につかえないかな〜
25
26
1 of 26

Recommended

実装レベルで学ぶVQVAE by
実装レベルで学ぶVQVAE実装レベルで学ぶVQVAE
実装レベルで学ぶVQVAEぱんいち すみもと
12.6K views23 slides
PRML学習者から入る深層生成モデル入門 by
PRML学習者から入る深層生成モデル入門PRML学習者から入る深層生成モデル入門
PRML学習者から入る深層生成モデル入門tmtm otm
5.7K views73 slides
[DL輪読会]Flow-based Deep Generative Models by
[DL輪読会]Flow-based Deep Generative Models[DL輪読会]Flow-based Deep Generative Models
[DL輪読会]Flow-based Deep Generative ModelsDeep Learning JP
14.6K views62 slides
GAN(と強化学習との関係) by
GAN(と強化学習との関係)GAN(と強化学習との関係)
GAN(と強化学習との関係)Masahiro Suzuki
83K views77 slides
[DL輪読会]NVAE: A Deep Hierarchical Variational Autoencoder by
[DL輪読会]NVAE: A Deep Hierarchical Variational Autoencoder[DL輪読会]NVAE: A Deep Hierarchical Variational Autoencoder
[DL輪読会]NVAE: A Deep Hierarchical Variational AutoencoderDeep Learning JP
1.9K views22 slides
【DL輪読会】A Time Series is Worth 64 Words: Long-term Forecasting with Transformers by
【DL輪読会】A Time Series is Worth 64 Words: Long-term Forecasting with Transformers【DL輪読会】A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
【DL輪読会】A Time Series is Worth 64 Words: Long-term Forecasting with TransformersDeep Learning JP
1.4K views21 slides

More Related Content

What's hot

深層生成モデルと世界モデル by
深層生成モデルと世界モデル深層生成モデルと世界モデル
深層生成モデルと世界モデルMasahiro Suzuki
16.6K views57 slides
Semi supervised, weakly-supervised, unsupervised, and active learning by
Semi supervised, weakly-supervised, unsupervised, and active learningSemi supervised, weakly-supervised, unsupervised, and active learning
Semi supervised, weakly-supervised, unsupervised, and active learningYusuke Uchida
6.3K views43 slides
畳み込みニューラルネットワークの高精度化と高速化 by
畳み込みニューラルネットワークの高精度化と高速化畳み込みニューラルネットワークの高精度化と高速化
畳み込みニューラルネットワークの高精度化と高速化Yusuke Uchida
64.5K views133 slides
自己教師学習(Self-Supervised Learning) by
自己教師学習(Self-Supervised Learning)自己教師学習(Self-Supervised Learning)
自己教師学習(Self-Supervised Learning)cvpaper. challenge
12.8K views177 slides
Generating Diverse High-Fidelity Images with VQ-VAE-2 by
Generating Diverse High-Fidelity Images with VQ-VAE-2Generating Diverse High-Fidelity Images with VQ-VAE-2
Generating Diverse High-Fidelity Images with VQ-VAE-2harmonylab
14.1K views21 slides
方策勾配型強化学習の基礎と応用 by
方策勾配型強化学習の基礎と応用方策勾配型強化学習の基礎と応用
方策勾配型強化学習の基礎と応用Ryo Iwaki
18.1K views42 slides

What's hot(20)

深層生成モデルと世界モデル by Masahiro Suzuki
深層生成モデルと世界モデル深層生成モデルと世界モデル
深層生成モデルと世界モデル
Masahiro Suzuki16.6K views
Semi supervised, weakly-supervised, unsupervised, and active learning by Yusuke Uchida
Semi supervised, weakly-supervised, unsupervised, and active learningSemi supervised, weakly-supervised, unsupervised, and active learning
Semi supervised, weakly-supervised, unsupervised, and active learning
Yusuke Uchida6.3K views
畳み込みニューラルネットワークの高精度化と高速化 by Yusuke Uchida
畳み込みニューラルネットワークの高精度化と高速化畳み込みニューラルネットワークの高精度化と高速化
畳み込みニューラルネットワークの高精度化と高速化
Yusuke Uchida64.5K views
自己教師学習(Self-Supervised Learning) by cvpaper. challenge
自己教師学習(Self-Supervised Learning)自己教師学習(Self-Supervised Learning)
自己教師学習(Self-Supervised Learning)
cvpaper. challenge12.8K views
Generating Diverse High-Fidelity Images with VQ-VAE-2 by harmonylab
Generating Diverse High-Fidelity Images with VQ-VAE-2Generating Diverse High-Fidelity Images with VQ-VAE-2
Generating Diverse High-Fidelity Images with VQ-VAE-2
harmonylab14.1K views
方策勾配型強化学習の基礎と応用 by Ryo Iwaki
方策勾配型強化学習の基礎と応用方策勾配型強化学習の基礎と応用
方策勾配型強化学習の基礎と応用
Ryo Iwaki18.1K views
【DL輪読会】The Forward-Forward Algorithm: Some Preliminary by Deep Learning JP
【DL輪読会】The Forward-Forward Algorithm: Some Preliminary【DL輪読会】The Forward-Forward Algorithm: Some Preliminary
【DL輪読会】The Forward-Forward Algorithm: Some Preliminary
Deep Learning JP2.2K views
【DL輪読会】High-Resolution Image Synthesis with Latent Diffusion Models by Deep Learning JP
【DL輪読会】High-Resolution Image Synthesis with Latent Diffusion Models【DL輪読会】High-Resolution Image Synthesis with Latent Diffusion Models
【DL輪読会】High-Resolution Image Synthesis with Latent Diffusion Models
Deep Learning JP2.6K views
[DLHacks]StyleGANとBigGANのStyle mixing, morphing by Deep Learning JP
[DLHacks]StyleGANとBigGANのStyle mixing, morphing[DLHacks]StyleGANとBigGANのStyle mixing, morphing
[DLHacks]StyleGANとBigGANのStyle mixing, morphing
Deep Learning JP14.2K views
【DL輪読会】Perceiver io a general architecture for structured inputs & outputs by Deep Learning JP
【DL輪読会】Perceiver io  a general architecture for structured inputs & outputs 【DL輪読会】Perceiver io  a general architecture for structured inputs & outputs
【DL輪読会】Perceiver io a general architecture for structured inputs & outputs
Deep Learning JP1.5K views
深層学習の不確実性 - Uncertainty in Deep Neural Networks - by tmtm otm
深層学習の不確実性 - Uncertainty in Deep Neural Networks -深層学習の不確実性 - Uncertainty in Deep Neural Networks -
深層学習の不確実性 - Uncertainty in Deep Neural Networks -
tmtm otm11.7K views
【DL輪読会】ViT + Self Supervised Learningまとめ by Deep Learning JP
【DL輪読会】ViT + Self Supervised Learningまとめ【DL輪読会】ViT + Self Supervised Learningまとめ
【DL輪読会】ViT + Self Supervised Learningまとめ
Deep Learning JP4K views
[DL輪読会]MetaFormer is Actually What You Need for Vision by Deep Learning JP
[DL輪読会]MetaFormer is Actually What You Need for Vision[DL輪読会]MetaFormer is Actually What You Need for Vision
[DL輪読会]MetaFormer is Actually What You Need for Vision
Deep Learning JP1.4K views
[DL輪読会]GLIDE: Guided Language to Image Diffusion for Generation and Editing by Deep Learning JP
[DL輪読会]GLIDE: Guided Language to Image Diffusion  for Generation and Editing[DL輪読会]GLIDE: Guided Language to Image Diffusion  for Generation and Editing
[DL輪読会]GLIDE: Guided Language to Image Diffusion for Generation and Editing
Deep Learning JP3K views
モデル高速化百選 by Yusuke Uchida
モデル高速化百選モデル高速化百選
モデル高速化百選
Yusuke Uchida24.7K views
【メタサーベイ】数式ドリブン教師あり学習 by cvpaper. challenge
【メタサーベイ】数式ドリブン教師あり学習【メタサーベイ】数式ドリブン教師あり学習
【メタサーベイ】数式ドリブン教師あり学習
cvpaper. challenge5.9K views
全力解説!Transformer by Arithmer Inc.
全力解説!Transformer全力解説!Transformer
全力解説!Transformer
Arithmer Inc.9.5K views
機械学習モデルの判断根拠の説明(Ver.2) by Satoshi Hara
機械学習モデルの判断根拠の説明(Ver.2)機械学習モデルの判断根拠の説明(Ver.2)
機械学習モデルの判断根拠の説明(Ver.2)
Satoshi Hara47.9K views
【DL輪読会】Efficiently Modeling Long Sequences with Structured State Spaces by Deep Learning JP
【DL輪読会】Efficiently Modeling Long Sequences with Structured State Spaces【DL輪読会】Efficiently Modeling Long Sequences with Structured State Spaces
【DL輪読会】Efficiently Modeling Long Sequences with Structured State Spaces
Deep Learning JP4.8K views
深層生成モデルと世界モデル(2020/11/20版) by Masahiro Suzuki
深層生成モデルと世界モデル(2020/11/20版)深層生成モデルと世界モデル(2020/11/20版)
深層生成モデルと世界モデル(2020/11/20版)
Masahiro Suzuki7.1K views

Viewers also liked

React.js + Flux入門 #scripty02 by
React.js + Flux入門 #scripty02React.js + Flux入門 #scripty02
React.js + Flux入門 #scripty02Yahoo!デベロッパーネットワーク
14.5K views71 slides
Web開発初心者がReactをチームに導入して半年経った by
Web開発初心者がReactをチームに導入して半年経ったWeb開発初心者がReactをチームに導入して半年経った
Web開発初心者がReactをチームに導入して半年経ったkazuki matsumura
687 views16 slides
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks by
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural NetworksDeep Learning JP
3.3K views12 slides
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo... by
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...Deep Learning JP
1.5K views38 slides
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images by
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People ImagesDeep Learning JP
1.6K views18 slides
[DLHacks] DLHacks説明資料 by
[DLHacks] DLHacks説明資料[DLHacks] DLHacks説明資料
[DLHacks] DLHacks説明資料Deep Learning JP
22.5K views11 slides

Viewers also liked(17)

Web開発初心者がReactをチームに導入して半年経った by kazuki matsumura
Web開発初心者がReactをチームに導入して半年経ったWeb開発初心者がReactをチームに導入して半年経った
Web開発初心者がReactをチームに導入して半年経った
kazuki matsumura687 views
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks by Deep Learning JP
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks
[DLHacks 実装] DeepPose: Human Pose Estimation via Deep Neural Networks
Deep Learning JP3.3K views
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo... by Deep Learning JP
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...
[DL輪読会] Towards an Automatic Turing Test: Learning to Evaluate Dialogue Respo...
Deep Learning JP1.5K views
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images by Deep Learning JP
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images
[DL輪読会] The Conditional Analogy GAN: Swapping Fashion Articles on People Images
Deep Learning JP1.6K views
[DLHacks 実装] The statistical recurrent unit by Deep Learning JP
[DLHacks 実装] The statistical recurrent unit[DLHacks 実装] The statistical recurrent unit
[DLHacks 実装] The statistical recurrent unit
Deep Learning JP974 views
[DL輪読会]Training RNNs as Fast as CNNs by Deep Learning JP
[DL輪読会]Training RNNs as Fast as CNNs[DL輪読会]Training RNNs as Fast as CNNs
[DL輪読会]Training RNNs as Fast as CNNs
Deep Learning JP1.8K views
[DL輪読会]Parallel Multiscale Autoregressive Density Estimation by Deep Learning JP
[DL輪読会]Parallel Multiscale Autoregressive Density Estimation[DL輪読会]Parallel Multiscale Autoregressive Density Estimation
[DL輪読会]Parallel Multiscale Autoregressive Density Estimation
Deep Learning JP728 views
[DLHacks 実装]Neural Machine Translation in Linear Time by Deep Learning JP
[DLHacks 実装]Neural Machine Translation in Linear Time [DLHacks 実装]Neural Machine Translation in Linear Time
[DLHacks 実装]Neural Machine Translation in Linear Time
Deep Learning JP1.6K views
[DL輪読会] DeepNav: Learning to Navigate Large Cities by Deep Learning JP
[DL輪読会] DeepNav: Learning to Navigate Large Cities[DL輪読会] DeepNav: Learning to Navigate Large Cities
[DL輪読会] DeepNav: Learning to Navigate Large Cities
Deep Learning JP1.1K views
[DLHacks 実装]Network Dissection: Quantifying Interpretability of Deep Visual R... by Deep Learning JP
[DLHacks 実装]Network Dissection: Quantifying Interpretability of Deep Visual R...[DLHacks 実装]Network Dissection: Quantifying Interpretability of Deep Visual R...
[DLHacks 実装]Network Dissection: Quantifying Interpretability of Deep Visual R...
Deep Learning JP1.3K views
[DL輪読会]Opening the Black Box of Deep Neural Networks via Information by Deep Learning JP
[DL輪読会]Opening the Black Box of Deep Neural Networks via Information[DL輪読会]Opening the Black Box of Deep Neural Networks via Information
[DL輪読会]Opening the Black Box of Deep Neural Networks via Information
Deep Learning JP1.7K views
[DLHacks LT] PytorchのDataLoader -torchtextのソースコードを読んでみた- by Deep Learning JP
[DLHacks LT] PytorchのDataLoader -torchtextのソースコードを読んでみた-[DLHacks LT] PytorchのDataLoader -torchtextのソースコードを読んでみた-
[DLHacks LT] PytorchのDataLoader -torchtextのソースコードを読んでみた-
Deep Learning JP36.2K views
[DLHacks 実装]Perceptual Adversarial Networks for Image-to-Image Transformation by Deep Learning JP
[DLHacks 実装]Perceptual Adversarial Networks for Image-to-Image Transformation[DLHacks 実装]Perceptual Adversarial Networks for Image-to-Image Transformation
[DLHacks 実装]Perceptual Adversarial Networks for Image-to-Image Transformation
Deep Learning JP2.1K views
[DL輪読会] MoCoGAN: Decomposing Motion and Content for Video Generation by Deep Learning JP
[DL輪読会] MoCoGAN: Decomposing Motion and Content for Video Generation[DL輪読会] MoCoGAN: Decomposing Motion and Content for Video Generation
[DL輪読会] MoCoGAN: Decomposing Motion and Content for Video Generation
Deep Learning JP2.9K views
[DL輪読会]Energy-based generative adversarial networks by Deep Learning JP
[DL輪読会]Energy-based generative adversarial networks[DL輪読会]Energy-based generative adversarial networks
[DL輪読会]Energy-based generative adversarial networks
Deep Learning JP1.8K views

Similar to [DL輪読会] Spectral Norm Regularization for Improving the Generalizability of Deep Learning/Spectral Normalization for GANs

[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる− by
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−Deep Learning JP
2.4K views23 slides
Diet networks thin parameters for fat genomic by
Diet networks thin parameters for fat genomicDiet networks thin parameters for fat genomic
Diet networks thin parameters for fat genomicHakky St
2.1K views36 slides
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい by
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたいTakuma Wakamori
1.6K views15 slides
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS by
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKSDeep Learning JP
4.3K views22 slides
attention_is_all_you_need_nips17_論文紹介 by
attention_is_all_you_need_nips17_論文紹介attention_is_all_you_need_nips17_論文紹介
attention_is_all_you_need_nips17_論文紹介Masayoshi Kondo
1.6K views42 slides
Dilated rnn by
Dilated rnnDilated rnn
Dilated rnnnaoto moriyama
3.1K views23 slides

Similar to [DL輪読会] Spectral Norm Regularization for Improving the Generalizability of Deep Learning/Spectral Normalization for GANs(20)

[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる− by Deep Learning JP
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
Deep Learning JP2.4K views
Diet networks thin parameters for fat genomic by Hakky St
Diet networks thin parameters for fat genomicDiet networks thin parameters for fat genomic
Diet networks thin parameters for fat genomic
Hakky St2.1K views
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい by Takuma Wakamori
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい
強化学習初心者が強化学習でニューラルネットワークの設計を自動化してみたい
Takuma Wakamori1.6K views
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS by Deep Learning JP
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS
[DL輪読会]QUASI-RECURRENT NEURAL NETWORKS
Deep Learning JP4.3K views
attention_is_all_you_need_nips17_論文紹介 by Masayoshi Kondo
attention_is_all_you_need_nips17_論文紹介attention_is_all_you_need_nips17_論文紹介
attention_is_all_you_need_nips17_論文紹介
Masayoshi Kondo1.6K views
Tree-to-Sequence Attentional Neural Machine Translation (ACL 2016) by Toru Fujino
Tree-to-Sequence Attentional Neural Machine Translation (ACL 2016)Tree-to-Sequence Attentional Neural Machine Translation (ACL 2016)
Tree-to-Sequence Attentional Neural Machine Translation (ACL 2016)
Toru Fujino535 views
LCCC2010:Learning on Cores, Clusters and Cloudsの解説 by Preferred Networks
LCCC2010:Learning on Cores,  Clusters and Cloudsの解説LCCC2010:Learning on Cores,  Clusters and Cloudsの解説
LCCC2010:Learning on Cores, Clusters and Cloudsの解説
Preferred Networks6.3K views
PFP:材料探索のための汎用Neural Network Potential_中郷_20220422POLセミナー by Matlantis
PFP:材料探索のための汎用Neural Network Potential_中郷_20220422POLセミナーPFP:材料探索のための汎用Neural Network Potential_中郷_20220422POLセミナー
PFP:材料探索のための汎用Neural Network Potential_中郷_20220422POLセミナー
Matlantis 1K views
Combinatorial optimization with graph convolutional networks and guided by Shuntaro Ohno
Combinatorial optimization with graph convolutional networks and guidedCombinatorial optimization with graph convolutional networks and guided
Combinatorial optimization with graph convolutional networks and guided
Shuntaro Ohno869 views
Combinatorial optimization with graph convolutional networks and guided ver20... by Shuntaro Ohno
Combinatorial optimization with graph convolutional networks and guided ver20...Combinatorial optimization with graph convolutional networks and guided ver20...
Combinatorial optimization with graph convolutional networks and guided ver20...
Shuntaro Ohno688 views
(文献紹介)Deep Unrolling: Learned ISTA (LISTA) by Morpho, Inc.
(文献紹介)Deep Unrolling: Learned ISTA (LISTA)(文献紹介)Deep Unrolling: Learned ISTA (LISTA)
(文献紹介)Deep Unrolling: Learned ISTA (LISTA)
Morpho, Inc.6.3K views
Image net classification with Deep Convolutional Neural Networks by Shingo Horiuchi
Image net classification with Deep Convolutional Neural NetworksImage net classification with Deep Convolutional Neural Networks
Image net classification with Deep Convolutional Neural Networks
Shingo Horiuchi9.1K views
Active Learning 入門 by Shuyo Nakatani
Active Learning 入門Active Learning 入門
Active Learning 入門
Shuyo Nakatani51.8K views
関西CVPRML勉強会資料20150627 by tsunekawas
関西CVPRML勉強会資料20150627関西CVPRML勉強会資料20150627
関西CVPRML勉強会資料20150627
tsunekawas1.3K views
Using Deep Learning for Recommendation by Eduardo Gonzalez
Using Deep Learning for RecommendationUsing Deep Learning for Recommendation
Using Deep Learning for Recommendation
Eduardo Gonzalez1.1K views
深層学習(岡本孝之 著) - Deep Learning chap.1 and 2 by Masayoshi Kondo
深層学習(岡本孝之 著) - Deep Learning chap.1 and 2深層学習(岡本孝之 著) - Deep Learning chap.1 and 2
深層学習(岡本孝之 著) - Deep Learning chap.1 and 2
Masayoshi Kondo1.4K views
Deep learning勉強会20121214ochi by Ohsawa Goodfellow
Deep learning勉強会20121214ochiDeep learning勉強会20121214ochi
Deep learning勉強会20121214ochi
Ohsawa Goodfellow26.8K views
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks... by Deep Learning JP
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
Deep Learning JP1.1K views

More from Deep Learning JP

【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners by
【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners
【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving PlannersDeep Learning JP
236 views28 slides
【DL輪読会】事前学習用データセットについて by
【DL輪読会】事前学習用データセットについて【DL輪読会】事前学習用データセットについて
【DL輪読会】事前学習用データセットについてDeep Learning JP
273 views20 slides
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP... by
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...Deep Learning JP
185 views26 slides
【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition by
【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition
【DL輪読会】AnyLoc: Towards Universal Visual Place RecognitionDeep Learning JP
255 views30 slides
【DL輪読会】Can Neural Network Memorization Be Localized? by
【DL輪読会】Can Neural Network Memorization Be Localized?【DL輪読会】Can Neural Network Memorization Be Localized?
【DL輪読会】Can Neural Network Memorization Be Localized?Deep Learning JP
515 views15 slides
【DL輪読会】Hopfield network 関連研究について by
【DL輪読会】Hopfield network 関連研究について【DL輪読会】Hopfield network 関連研究について
【DL輪読会】Hopfield network 関連研究についてDeep Learning JP
1.3K views29 slides

More from Deep Learning JP(20)

【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners by Deep Learning JP
【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners
【DL輪読会】AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners
Deep Learning JP236 views
【DL輪読会】事前学習用データセットについて by Deep Learning JP
【DL輪読会】事前学習用データセットについて【DL輪読会】事前学習用データセットについて
【DL輪読会】事前学習用データセットについて
Deep Learning JP273 views
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP... by Deep Learning JP
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...
【DL輪読会】 "Learning to render novel views from wide-baseline stereo pairs." CVP...
Deep Learning JP185 views
【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition by Deep Learning JP
【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition
【DL輪読会】AnyLoc: Towards Universal Visual Place Recognition
Deep Learning JP255 views
【DL輪読会】Can Neural Network Memorization Be Localized? by Deep Learning JP
【DL輪読会】Can Neural Network Memorization Be Localized?【DL輪読会】Can Neural Network Memorization Be Localized?
【DL輪読会】Can Neural Network Memorization Be Localized?
Deep Learning JP515 views
【DL輪読会】Hopfield network 関連研究について by Deep Learning JP
【DL輪読会】Hopfield network 関連研究について【DL輪読会】Hopfield network 関連研究について
【DL輪読会】Hopfield network 関連研究について
Deep Learning JP1.3K views
【DL輪読会】SimPer: Simple self-supervised learning of periodic targets( ICLR 2023 ) by Deep Learning JP
【DL輪読会】SimPer: Simple self-supervised learning of periodic targets( ICLR 2023 )【DL輪読会】SimPer: Simple self-supervised learning of periodic targets( ICLR 2023 )
【DL輪読会】SimPer: Simple self-supervised learning of periodic targets( ICLR 2023 )
Deep Learning JP335 views
【DL輪読会】RLCD: Reinforcement Learning from Contrast Distillation for Language M... by Deep Learning JP
【DL輪読会】RLCD: Reinforcement Learning from Contrast Distillation for Language M...【DL輪読会】RLCD: Reinforcement Learning from Contrast Distillation for Language M...
【DL輪読会】RLCD: Reinforcement Learning from Contrast Distillation for Language M...
Deep Learning JP233 views
【DL輪読会】"Secrets of RLHF in Large Language Models Part I: PPO" by Deep Learning JP
【DL輪読会】"Secrets of RLHF in Large Language Models Part I: PPO"【DL輪読会】"Secrets of RLHF in Large Language Models Part I: PPO"
【DL輪読会】"Secrets of RLHF in Large Language Models Part I: PPO"
Deep Learning JP776 views
【DL輪読会】"Language Instructed Reinforcement Learning for Human-AI Coordination " by Deep Learning JP
【DL輪読会】"Language Instructed Reinforcement Learning  for Human-AI Coordination "【DL輪読会】"Language Instructed Reinforcement Learning  for Human-AI Coordination "
【DL輪読会】"Language Instructed Reinforcement Learning for Human-AI Coordination "
Deep Learning JP446 views
【DL輪読会】Llama 2: Open Foundation and Fine-Tuned Chat Models by Deep Learning JP
【DL輪読会】Llama 2: Open Foundation and Fine-Tuned Chat Models【DL輪読会】Llama 2: Open Foundation and Fine-Tuned Chat Models
【DL輪読会】Llama 2: Open Foundation and Fine-Tuned Chat Models
Deep Learning JP1.3K views
【DL輪読会】"Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware" by Deep Learning JP
【DL輪読会】"Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware"【DL輪読会】"Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware"
【DL輪読会】"Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware"
Deep Learning JP415 views
【DL輪読会】Parameter is Not All You Need:Starting from Non-Parametric Networks fo... by Deep Learning JP
【DL輪読会】Parameter is Not All You Need:Starting from Non-Parametric Networks fo...【DL輪読会】Parameter is Not All You Need:Starting from Non-Parametric Networks fo...
【DL輪読会】Parameter is Not All You Need:Starting from Non-Parametric Networks fo...
Deep Learning JP398 views
【DL輪読会】Drag Your GAN: Interactive Point-based Manipulation on the Generative ... by Deep Learning JP
【DL輪読会】Drag Your GAN: Interactive Point-based Manipulation on the Generative ...【DL輪読会】Drag Your GAN: Interactive Point-based Manipulation on the Generative ...
【DL輪読会】Drag Your GAN: Interactive Point-based Manipulation on the Generative ...
Deep Learning JP687 views
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive... by Deep Learning JP
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...
Deep Learning JP816 views
【DL輪読会】Towards Understanding Ensemble, Knowledge Distillation and Self-Distil... by Deep Learning JP
【DL輪読会】Towards Understanding Ensemble, Knowledge Distillation and Self-Distil...【DL輪読会】Towards Understanding Ensemble, Knowledge Distillation and Self-Distil...
【DL輪読会】Towards Understanding Ensemble, Knowledge Distillation and Self-Distil...
Deep Learning JP376 views
【DL輪読会】Deep Transformers without Shortcuts: Modifying Self-attention for Fait... by Deep Learning JP
【DL輪読会】Deep Transformers without Shortcuts: Modifying Self-attention for Fait...【DL輪読会】Deep Transformers without Shortcuts: Modifying Self-attention for Fait...
【DL輪読会】Deep Transformers without Shortcuts: Modifying Self-attention for Fait...
Deep Learning JP329 views
【DL輪読会】マルチモーダル 基盤モデル by Deep Learning JP
【DL輪読会】マルチモーダル 基盤モデル【DL輪読会】マルチモーダル 基盤モデル
【DL輪読会】マルチモーダル 基盤モデル
Deep Learning JP1.1K views
【DL輪読会】TrOCR: Transformer-based Optical Character Recognition with Pre-traine... by Deep Learning JP
【DL輪読会】TrOCR: Transformer-based Optical Character Recognition with Pre-traine...【DL輪読会】TrOCR: Transformer-based Optical Character Recognition with Pre-traine...
【DL輪読会】TrOCR: Transformer-based Optical Character Recognition with Pre-traine...
Deep Learning JP722 views
【DL輪読会】HyperDiffusion: Generating Implicit Neural Fields withWeight-Space Dif... by Deep Learning JP
【DL輪読会】HyperDiffusion: Generating Implicit Neural Fields withWeight-Space Dif...【DL輪読会】HyperDiffusion: Generating Implicit Neural Fields withWeight-Space Dif...
【DL輪読会】HyperDiffusion: Generating Implicit Neural Fields withWeight-Space Dif...
Deep Learning JP246 views

Recently uploaded

Keycloakの全体像: 基本概念、ユースケース、そして最新の開発動向 by
Keycloakの全体像: 基本概念、ユースケース、そして最新の開発動向Keycloakの全体像: 基本概念、ユースケース、そして最新の開発動向
Keycloakの全体像: 基本概念、ユースケース、そして最新の開発動向Hitachi, Ltd. OSS Solution Center.
60 views26 slides
The Things Stack説明資料 by The Things Industries by
The Things Stack説明資料 by The Things IndustriesThe Things Stack説明資料 by The Things Industries
The Things Stack説明資料 by The Things IndustriesCRI Japan, Inc.
58 views29 slides
さくらのひやおろし2023 by
さくらのひやおろし2023さくらのひやおろし2023
さくらのひやおろし2023法林浩之
96 views58 slides
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料) by
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)NTT DATA Technology & Innovation
18 views38 slides
Windows 11 information that can be used at the development site by
Windows 11 information that can be used at the development siteWindows 11 information that can be used at the development site
Windows 11 information that can be used at the development siteAtomu Hidaka
88 views41 slides

Recently uploaded(11)

The Things Stack説明資料 by The Things Industries by CRI Japan, Inc.
The Things Stack説明資料 by The Things IndustriesThe Things Stack説明資料 by The Things Industries
The Things Stack説明資料 by The Things Industries
CRI Japan, Inc.58 views
さくらのひやおろし2023 by 法林浩之
さくらのひやおろし2023さくらのひやおろし2023
さくらのひやおろし2023
法林浩之96 views
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料) by NTT DATA Technology & Innovation
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)
速習! PostgreSQL専用HAソフトウェア: Patroni(PostgreSQL Conference Japan 2023 発表資料)
Windows 11 information that can be used at the development site by Atomu Hidaka
Windows 11 information that can be used at the development siteWindows 11 information that can be used at the development site
Windows 11 information that can be used at the development site
Atomu Hidaka88 views
今、改めて考えるPostgreSQLプラットフォーム - マルチクラウドとポータビリティ -(PostgreSQL Conference Japan 20... by NTT DATA Technology & Innovation
今、改めて考えるPostgreSQLプラットフォーム - マルチクラウドとポータビリティ -(PostgreSQL Conference Japan 20...今、改めて考えるPostgreSQLプラットフォーム - マルチクラウドとポータビリティ -(PostgreSQL Conference Japan 20...
今、改めて考えるPostgreSQLプラットフォーム - マルチクラウドとポータビリティ -(PostgreSQL Conference Japan 20...
SNMPセキュリティ超入門 by mkoda
SNMPセキュリティ超入門SNMPセキュリティ超入門
SNMPセキュリティ超入門
mkoda355 views
SSH応用編_20231129.pdf by icebreaker4
SSH応用編_20231129.pdfSSH応用編_20231129.pdf
SSH応用編_20231129.pdf
icebreaker4335 views

[DL輪読会] Spectral Norm Regularization for Improving the Generalizability of Deep Learning/Spectral Normalization for GANs