技術動向の調査として、ICML Workshop Uncertainty & Robustness in Deep Learningの中で、面白そうなタイトルを中心に読んで各論文を4スライドでまとめました。
最新版:https://speakerdeck.com/masatoto/icml-2021-workshop-shen-ceng-xue-xi-falsebu-que-shi-xing-nituite-e0debbd2-62a7-4922-a809-cb07c5da2d08(文章を修正しました。)
The document discusses control as inference in Markov decision processes (MDPs) and partially observable MDPs (POMDPs). It introduces optimality variables that represent whether a state-action pair is optimal or not. It formulates the optimal action-value function Q* and optimal value function V* in terms of these optimality variables and the reward and transition distributions. Q* is defined as the log probability of a state-action pair being optimal, and V* is defined as the log probability of a state being optimal. Bellman equations are derived relating Q* and V* to the reward and next state value.
[DL輪読会]Recent Advances in Autoencoder-Based Representation LearningDeep Learning JP
?
1. Recent advances in autoencoder-based representation learning include incorporating meta-priors to encourage disentanglement and using rate-distortion and rate-distortion-usefulness tradeoffs to balance compression and reconstruction.
2. Variational autoencoders introduce priors to disentangle latent factors, but recent work aggregates posteriors to directly encourage disentanglement.
3. The rate-distortion framework balances the rate of information transmission against reconstruction distortion, while rate-distortion-usefulness also considers downstream task usefulness.
The document discusses control as inference in Markov decision processes (MDPs) and partially observable MDPs (POMDPs). It introduces optimality variables that represent whether a state-action pair is optimal or not. It formulates the optimal action-value function Q* and optimal value function V* in terms of these optimality variables and the reward and transition distributions. Q* is defined as the log probability of a state-action pair being optimal, and V* is defined as the log probability of a state being optimal. Bellman equations are derived relating Q* and V* to the reward and next state value.
[DL輪読会]Recent Advances in Autoencoder-Based Representation LearningDeep Learning JP
?
1. Recent advances in autoencoder-based representation learning include incorporating meta-priors to encourage disentanglement and using rate-distortion and rate-distortion-usefulness tradeoffs to balance compression and reconstruction.
2. Variational autoencoders introduce priors to disentangle latent factors, but recent work aggregates posteriors to directly encourage disentanglement.
3. The rate-distortion framework balances the rate of information transmission against reconstruction distortion, while rate-distortion-usefulness also considers downstream task usefulness.
ICML2013読み会 Large-Scale Learning with Less RAM via RandomizationHidekazu Oiwa
?
Large-Scale Learning with Less RAM via Randomization proposes algorithms that reduce memory usage for machine learning models during training and prediction while maintaining prediction accuracy. It introduces a method called randomized rounding that represents model weights with fewer bits by randomly rounding values to the nearest representation. An algorithm is proposed that uses randomized rounding and adaptive learning rates on a per-coordinate basis, providing theoretical guarantees on regret bounds. Memory usage is reduced by 50% during training and 95% during prediction compared to standard floating point representation.
The document discusses the EM algorithm and K-means clustering. It begins by introducing mixture models and the EM algorithm for parameter estimation. It then describes the K-means clustering algorithm and how it can be viewed as a special case of EM. The document concludes by explaining how EM can be applied to Gaussian mixture models, deriving the E and M steps, and introducing the responsibilities that indicate cluster assignments.
Protect Your IoT Data with UbiBot's Private Platform.pptxユビボット 株式会社
?
Our on-premise IoT platform offers a secure and scalable solution for businesses, with features such as real-time monitoring, customizable alerts and open API support, and can be deployed on your own servers to ensure complete data privacy and control.
21. Importance-aware step width
ステップ幅の再設定式
Table 1: Importance Weight Aware Updates for Various Loss Functions
Loss
`(p, y)
Update s(h)
?
?
>
p y
Squared
(y p)2
1 e h?x x
x> x
Logistic
log(1 + e
Exponential
e
y log
Logarithmic
Hellinger
Hinge
? -Quantile
p
( p
y
p
p
2
y)
yp
)
yp
+ (1
p
( 1
y) log
p
1 y
1 p
p
1
max(0, 1 yp)
if y > p
? (y p)
if y ? p (1 ? )(p y)
(6) gives a di?erential equation whose solution is the
result of a continuous gradient descent process.
As a sanity check we rederive (5) using (6). For
@`
squared loss @p = p y and we get a linear ODE:
y)2
> x+yp+eyp
) h?x> x eyp
for y 2 { 1, 1}
yx> x
py log(h?x> x+epy )
for y 2 { 1, 1}
x> xy
p
p 1+ (p 1)2 +2h?x> x
if y = 0
p x> x
p
p2 +2h?x> x
if y = 1
x> x
>
1
p 1+ 4 (12h?x x+8(1 p)3/2 )2/3
if y = 0
x> x
1
p 4 (12h?x> x+8p3/2 )2/3
if y = 1
x> x
1 yp
y min h?, x> x for y 2 { 1, 1}
if y > p
? min(h?, ?yx>p )
x
p y
if y ? p (1 ? ) min(h?, (1 ? )x> x )
W (eh?x
solution to (6) has no simple form for all y 2 [0, 1] but
for y 2 {0, 1} we get the expressions in table 1.
3.1.1
(Karampatziakis+ 11) より
Hinge Loss and Quantile Loss
Two other commonly used loss function are the hinge
loss
21 and the ? -quantile loss where ? 2 [0, 1] is a parameter function. These are di?erentiable everywhere
31. Normalized Online Learning
? オンライン処理しながら自動で正規化
? スケールを(あまり)気にせず,SGDを回せるように!
? スケールも敵対的に設定されるRegret Boundの証明付き
Algorithm 1 NG(learning rate ?t )
Algorithm 2 NAG(learning rate ?)
1. Initially wi = 0, si = 0, N = 0
1. Initially wi = 0, si = 0, Gi = 0, N
2. For each timestep t observe example (x, y)
2. For each timestep t observe example
(a) For each i, if |xi | > si
(a) For each i, if |xi | > si
wi si
i. wi
|xi |
ii. si
|xi |
P
(b) y = i wi xi
?
P x2
i
(c) N
N + i s2
wi s2
i
|xi |2
i. wi
ii. si
|xi |
P
(b) y = i wi xi
?
P
(c) N
N+ i
(d) For each i,
i. wi
wi
x2
i
2
si
(d) For each i,
y ,y)
t
?t N s1 @L(?i
2
@w
i
31
i. Gi
Gi +
ii. wi
wi
i
?
@L(?,y)
y
@wi
(Stéphane+ 13)より q t
?
N si
?2
1
p
@L
Gi @