95 Matching Annotations
  1. Last 7 days
    1. 火星軍情局  · odsStpenro c6064g M:43Ahpt04tmr2 lc134f89A8l17a ilut98652g4g  · Shared with Public還在考慮要不要看騰訊版的《三體》嗎?和Netflix版哪個好?會這麼想的人真是時間太多,騰訊版是30集連續劇啊!而我看完了。呃~其實是1.75倍快轉再不時跳著看啦。這是個有趣的「一書各表」的體驗:一本快20年前的中國科幻小說,現在由東西方截然不同的文化來表述,兩者的差距頗大且極具意義。哪個版本好?那要看各人對“好”的定義。我比較喜歡Netflix,但能完全理解支持騰訊的觀點。下面是我對騰訊版抓的重點:1) 三十集連續劇呀!現在誰有這種美國~不~中國時間。2)對「原著黨」來說這是唯一選擇,大部分原封不動來自小說。3) 但這也是問題,有些地方書寫得很有趣,但是演出來就乏味。比如:小說中主角汪淼發現底片竟然在倒數計時的過程懸疑而有張力,但是演出來就很脫戲。主角照相、沖洗照片、嚇一跳、繼續照相、沖洗照片、嚇一跳、繼續照相…半集就沒了。4)看完Netflix以後再看騰訊版,會覺得騰訊虛擬世界的電腦動畫的品質欠佳,讓人出戲,畢竟騰訊的成本只是Netflix的零頭。(但Netflix未免太會花錢了!)5)騰訊版似乎沒有要求演技,每個演員都面無表情地念台詞,只有一個例外:警官大史,他一出現可看性就陡升一個數量級。6)騰訊版也並非完全遵照原著,小說裡中共的黑歷史都不見了。葉文潔的故事開始於響應知識青年下鄉而到大興安嶺,之前爸爸被鬥死的部分似乎沒發生。我不是愛黑中共,而是那段歷史對葉文潔的動機有很深的影響,沒那段就讓故事有些莫名其妙。7)還有一個例子:小說裡有一位法官去找苦牢裡冷得發抖的葉文潔,看似非常慈祥,勸葉文潔簽署一份控告葉父的文件,但葉文潔拒絕。騰訊版就演到這裡,但看過小說或Netflix版的人都知道後來法官大怒,倒桶冰水在葉文潔身上害她差點冷死。看騰訊版會覺得法官人很好,是葉文潔不識抬舉。

      看完Netflix第一季,也看過騰訊第一集,完全同意這裡的分析。

      火星軍情局 · 還在考慮要不要看騰訊版的《三體》嗎?和Netflix版哪個好?會這麼想的人真是時間太多,騰訊版是30集連續劇啊! 而我看完了。呃~其實是1.75倍快轉再不時跳著看啦。 這是個有趣的「一書各表」的體驗:一本快20年前的中國科幻小說,現在由東西方截然不同的文化來表述,兩者的差距頗大且極具意義。 哪個版本好?那要看各人對“好”的定義。我比較喜歡Netflix,但能完全理解支持騰訊的觀點。下面是我對騰訊版抓的重點: 1) 三十集連續劇呀!現在誰有這種美國~不~中國時間。 2)對「原著黨」來說這是唯一選擇,大部分原封不動來自小說。 3) 但這也是問題,有些地方書寫得很有趣,但是演出來就乏味。比如:小說中主角汪淼發現底片竟然在倒數計時的過程懸疑而有張力,但是演出來就很脫戲。主角照相、沖洗照片、嚇一跳、繼續照相、沖洗照片、嚇一跳、繼續照相…半集就沒了。 4)看完Netflix以後再看騰訊版,會覺得騰訊虛擬世界的電腦動畫的品質欠佳,讓人出戲,畢竟騰訊的成本只是Netflix的零頭。(但Netflix未免太會花錢了!) 5)騰訊版似乎沒有要求演技,每個演員都面無表情地念台詞,只有一個例外:警官大史,他一出現可看性就陡升一個數量級。 6)騰訊版也並非完全遵照原著,小說裡中共的黑歷史都不見了。葉文潔的故事開始於響應知識青年下鄉而到大興安嶺,之前爸爸被鬥死的部分似乎沒發生。我不是愛黑中共,而是那段歷史對葉文潔的動機有很深的影響,沒那段就讓故事有些莫名其妙。 7)還有一個例子:小說裡有一位法官去找苦牢裡冷得發抖的葉文潔,看似非常慈祥,勸葉文潔簽署一份控告葉父的文件,但葉文潔拒絕。騰訊版就演到這裡,但看過小說或Netflix版的人都知道後來法官大怒,倒桶冰水在葉文潔身上害她差點冷死。看騰訊版會覺得法官人很好,是葉文潔不識抬舉。

  2. Apr 2024
  3. Mar 2024
  4. Jan 2024
  5. Dec 2023
      • for: polycrisis, Thomas Homer-Dixon, Cascade Institute Royal Roads University - Changemakers Speakers Series, etymology - polylcrisis

      • Talk: Hope in the Polycrisis

      • Speaker: Thomas Homer-Dixon
      • Host: Royal Roads University - Changemakers Speakers Series
      • Date: 2023

      • SUMMARY

        • Thomas Homer-Dixon is a leading complex systems scientist and director of the Cascade Institute, which he co-founded at Royal Roads University in Victoria, B.C., Canada, to study the polycrisis and identify strategic high leverage interventions that could rapidly shift humanity's trajectory in the next few critical years.
        • The talk, entitled "Hope in the polycriisis" chronicles Homer-Dixon's multi-decade journey to understand the convergence of crisis happening in the world today.
        • In a real sense, the evolution of his thinking on these complex problems are reflected in the series of books he has written over the years, culminating in the 2023 book "Commanding Hope", based on a theory of hope:

          • Environment, Scarcity, and Violence (Princeton, 1999). - a book showing how other factors combine with environmental stress to produce violence.
          • “The Ingenuity Gap: Can Poor Countries Adapt to Resource Scarcity?,” which appeared in Population and Development Review in 1995
          • “Resource Scarcity and Innovation: Can Poor Countries Attain Endogenous Growth?" ?” coauthored with Edward Barbier, which appeared in Ambio (1999)
          • The Upside of Down: Catastrophe, Creativity, and the Renewal of Civilization (2006), examined the threat to global stability of simultaneous and interacting demographic, environmental, economic, and political stresses. This led to examining energy as a major factor in our modern society.
          • "Commanding Hope: The Power we have to Renew a World in Peril"
        • Homer-Dixon also talks about practical solutions, His team at Casacade Institute is researching a promising technology called ultra-deep geothermal, which could provide unlimted energy at energy densities comprable to fossil fuels.

        • He finishes his talk with his theory of Hope and how a "Robust" hope can be the key to a successful rapid transition.
      • etymology - polycrisis

        • https://polycrisis.org/lessons/where-did-the-term-polycrisis-come-from/
        • Complexity theorists Edgar Morin and Anne Brigitte Kern first used the term polycrisis in their 1999 book, Homeland Earth, to argue that the world faces
          • “no single vital problem, but many vital problems, and it is this complex intersolidarity of problems, antagonisms, crises, uncontrolled processes, and the general crisis of the planet that constitutes the number one vital problem" (p. 74).
        • South African sociologist and sustainable transitions theorist Mark Swilling then adopted the term to capture
          • “a nested set of globally interactive socio-economic, ecological and cultural-institutional crises that defy reduction to a single cause” (2013, p. 98).
        • Climate change, rising inequality, and the threat of financial crises interact in complex ways that multiply their overall impact (Swilling 2013, 2019).
  6. Sep 2023
  7. Jul 2023
  8. Mar 2023
  9. Nov 2022
    1. Dr. Miho Ohsaki re-examined workshe and her group had previously published and confirmed that the results are indeed meaningless in the sensedescribed in this work (Ohsaki et al., 2002). She has subsequently been able to redefine the clustering subroutine inher work to allow more meaningful pattern discovery (Ohsaki et al., 2003)

      Look into what Dr. Miho Ohsaki changed about the clustering subroutine in her work and how it allowed for "more meaningful pattern discovery"

    2. http://www.cs.ucr.edu/~eamonn/meaningless.pdf Paper that argues cluster time series subsequences is "meaningless". tl;dr: radically different distributions end up converging to translations of basic sine or trig functions. Wonder if constructing a simplicial complex does anything?

      Note that one researcher changed the algorithm to produce potentially meaningful results

  10. Oct 2022
  11. Sep 2022
  12. Jul 2022
    1. with established worldwide fame and prestige, to step in his previous successes to write more-of-the-same books and convert all the attention in cheap money. Just like Robert Kiyosaki did with his 942357 books about “Rich dad”.

      Many artists fall into a creativity trap caused by fame. They spend years developing a great work, but then when it's released, the industry requires they follow it up almost immediately with something even stronger.

      Jewel is an reasonable and perhaps typical example of this phenomenon. She spent several years writing the entirety of her first album Pieces of You (1995), which had three to four solid singles. As it became popular she was rushed to release Spirit (1998), which, while it was ultimately successful, didn't measure up to the first album which had far more incubation time. She wasn't able to build up enough material over time to more easily ride her initial wave of fame. Creativity on demand can be a difficult master, particularly when one is actively touring or supporting their first work while needing to

      (Compare the number of titles she self-wrote on album one vs. album two).

      M. Night Shyamalan is in a similar space, though as a director he preferred to direct scripts that he himself had written. While he'd had several years of writing and many scripts, some were owned by other production companies/studios which forced him to start from scratch several times and both write and direct at the same time, a process which is difficult to do by oneself.

      Another example is Robert Kiyosaki who spun off several similar "Rich Dad" books after the success of his first.

      Compare this with artists who have a note taking or commonplacing practice for maintaining the velocity of their creative output: - Eminem - stacking ammo - Taylor Swift - commonplace practice

  13. Jun 2022
    1. showRSS tracks most TV shows, so you don't have to. RSS integration lets you automate your set up.
  14. May 2022
  15. Jan 2022
  16. May 2021
  17. Mar 2021
  18. Dec 2020
    1. The ground-water compo- nent of streamflow was estimated from a streamflow hydrograph for the Homochitto River in Mississippi,

      pengamatan time series seperti ini jarang dilakukan dalam penelitian-penelitian hidrogeologi di indonesia.

  19. Oct 2020
  20. Sep 2020
  21. Aug 2020
  22. Jul 2020
  23. Jun 2020
    1. Kempfert, K., Martinez, K., Siraj, A., Conrad, J., Fairchild, G., Ziemann, A., Parikh, N., Osthus, D., Generous, N., Del Valle, S., & Manore, C. (2020). Time Series Methods and Ensemble Models to Nowcast Dengue at the State Level in Brazil. ArXiv:2006.02483 [q-Bio, Stat]. http://arxiv.org/abs/2006.02483

  24. May 2020
  25. Nov 2019
    1. This booklet itells you how to use the R statistical software to carry out some simple analyses that are common in analysing time series data.

      what is time series?

  26. Sep 2019
    1. Variation in Rates of Fatal Police Shootings across US States:the Role of Firearm Availability

      Hello! This article is about the relationship between firearm prevalence (the IV) and the rates of fatal police shootings (the DV). The authors hypothesize that the greater the prevalence of firearms, the grater the rates of fatal police shooting.

      This article follows the classical structure of social science research -- abstract, introduction, literature review/theory, research design, findings, and conclusions.

  27. Jul 2019
    1. It seems to me that in considering the purpose of a subset of the entire series of transactions, Justice Paris had in mind the correct test. As this Court said in MacKay v. Canada, 2008 FCA 105 (CanLII) at paragraph 25: The existence of a bona fide non-tax purpose for a series of transactions does not exclude the possibility that the primary purpose of one or more transactions within the series is to obtain a tax benefit.

      Existence of a bona fide non-tax purpose for a series of transactions does not exclude possibility that the primary purpose of one of the transactions in the series is to obtain a tax benefit.

  28. Jun 2019
    1. MelNet: A Generative Model for Audio in the Frequency Domain

      本文的主要贡献如下:

      • 提出了 MelNet。一个语谱图的生成模型,它结合了细粒度的自回归模型和多尺度生成过程,能够同时捕获局部和全局的结构。

      • 展示了 MelNet 在长程依赖性上卓越的性能。

      • 展示了 MelNet 在多种音频生成任务上优秀的能力:无条件语音生成任务、音乐生成任务、文字转语音合成任务。而且在这些任务上,MelNet 都是端到端的实现。

  29. Feb 2019
    1. 0 \0$aH

      The value in the original 490 is a 4XX in the provided series statement; it is incorrect, but that's not really Marcive's fault

  30. Jan 2019
    1. MAD-GAN: Multivariate Anomaly Detection for Time Series Data with Generative Adversarial Networks

      这 paper 挺神的,用 GAN 做时序数据异常检测。主要神在 G 和 D 都仅用 LSTM-RNN 来构造的!不仅因此值得我关注,更因为该模型可以为自己思考“非模板引力波探测”带来启发!

  31. Dec 2018
    1. Deep Neural Networks for Automatic Classification of Anesthetic-Induced Unconsciousness

      spatio-temporo-spectral features.

    2. Using Convolutional Neural Networks to Classify Audio Signal in Noisy Sound Scenes

      先辨别信号位置,再过滤出信号,这和 LIGO 找event波形的套路很像~ ;又看到 RNN与CNN 结合起来的应用~

    3. Sound Event Detection Using Spatial Features and Convolutional Recurrent Neural Network.

      输入数据是多通道音频信号,网络是结合了CNN 和 LSTM。

  32. Nov 2018
    1. Multilevel Wavelet Decomposition Network for Interpretable Time Series Analysis

      初步扫了一眼,感觉这篇文章应该可以给我一些 idea,内含我感兴趣(看得懂)的方法/机制,另外综述的参考文献对我来时也应该很有帮助。

      本文是北京航空航天大学发表于KDD 2018的文章,作者提出了将小波变换和深度神经网络进行完美结合,克服了融合的损失,对时间序列数据的分析起到了很好的启发性研究。

      Summary

    2. Foundations of Sequence-to-Sequence Modeling for Time Series

      利用序列到序列模型来做时序数据预测的理论研究 paper~

    3. Interpretable Convolutional Filters with SincNet

      一篇值得我高度关注的 paper,来自 AI 三巨头之一 Yoshua Bengio!其背后的核心是将数字信号处理DSP中卷积的激励函数(滤波器)进行了重新设计,不仅会保留了卷积的特性(线性性+时间平移不变性)还在滤波器上添加待学习参数来学习合适的高低频截断位置。

    4. A Deep Neural Network for Unsupervised Anomaly Detection and Diagnosis in Multivariate Time Series Data

      虽然数据特点为多变量时序信号+噪声少还频率低,不过作者提出的 Multi-Scale Convolutional Recurrent Encoder-Decoder (MSCRED) 网络很有趣,可见基于注意力机制的 ConvLSTM 在模式识别上是大为有用的!

      此外paper里的数学表述和实验讨论也很值得参考学习,算是非常标准的基于新model的 paper 样板~

    5. Deep Convolutional Neural Networks On Multichannel Time Series For Human Activity Recognition.

      这个文章的研究对象是 HAR 问题。不过这里的多通道时序信号是排成 长 x 宽维度,模拟图片数据来解决的,并不是图像的多通道。不过最后的效果貌似很不错,远胜过 SVM,KNN,MV,以及 DBN(深度置信网络)。

      文章对网络结构的英文描述还是值得借鉴的~

    6. Towards a universal neural network encoder for time series

      数据任务是“时序序列的分类”,这是我感兴趣的问题。Universal 代表不需要额外设置和训练,从某数据集训练后,就可以拿到另一个新训练集类型去搞事情~ 另一个特点是用了 encoder 得到了低维不变的表示。

    7. Time Series Classification Using Multi-Channels Deep Convolutional Neural Networks

      这是一个三通道并列输入的时序信号模型。效果貌似不错,还针对不同模型算法的预测时间在不同训练规模数据上的模型表现做了对比,

      另外文章的模型图示也很有启发性。

    8. Stochastic Adaptive Neural Architecture Search for Keyword Spotting

      一篇讲 identifying keywords in a real-time audio stream 的 paper。这和引力波探测中的数据处理很接近哦~!此文提出 end-end 的“随机自适应神经构架搜寻” (SANAS) 实现高效准确的训练效果。这显然对 real-time 特点的类型数据应用带来启发。FYI:人家源码还开放了。。。

    9. WaveGlow: A Flow-based Generative Network for Speech Synthesis

      一篇来自 NVIDIA 的小文。提出的实时生成网络 WaveGlow 结合了 Glow 和 WaveNet 的特点,实现了更快速高效准确的语音合成。

    10. Deep learning for time series classification: a review

      这个文与我的课题貌似相当相关!

      准备好好写一个 Paper Summary 为好~

    11. Deep Learning for Time-Series Analysis

      一个比较简洁的关于时序序列的 DL 应用的综述文章。

      其中也有谈论时序序列的分类问题。

    12. Whispered-to-voiced Alaryngeal Speech Conversion with Generative Adversarial Networks

      这是一篇用 GAN 来做 Voiced Speech Restoration 的,并且使用了作者自己提出的 speech enhancement using GANs (SEGAN) 。

      于我而言,亮点有二:

      1. 数据是时序语音
      2. 利用 GAN 对语音的增强效果似乎对降噪有些启发
      3. 网络结构图画的蛮好看的:

    13. Model Selection Techniques -- An Overview

      一篇关于模型选择的综述文章。涉及信号处理,图像处理等等多方面数据信息的处理。发表在信号处理的期刊杂志上。

      文中关于模型选择的大概念方向,和数学表示,是值得好好阅读的。

    14. End-to-end music source separation: is it possible in the waveform domain?

      讨论的是 Music source separation 问题。

      作者认为前人基于spectrogram的输入数据都忽略了相位信息,所以提出了直接waveform-based的模型得到了明显更好的效果。

    15. Unifying Probabilistic Models for Time-Frequency Analysis

      文章涉及好些自己还没搞懂的概念和方法:

      • 时频分析
      • Gaussian processes
      • ...

      文中的 review 给的是很不错的~

  33. Oct 2018
  34. Apr 2017
    1. Ye say that the interest of the master is a sufficient protection to the slave. In the fury of man’s mad will, he will wittingly and with open eyes sell his own soul to the Devil to get his ends; and will he be more careful of his neighbor’s body?

      Having a master's favor could mean a much better life, and since Legree hated Tom it meant that he wanted to kill him even though it likened him to sin.