linkstream2 microblog

https://www.nature.com/articles/s41586-019-0912-1
Perspective | Published: 13 February 2019
Deep learning and process understanding for data-driven Earth system science Markus Reichstein, Gustau Camps-Valls, Bjorn Stevens, Martin Jung, Joachim Denzler, Nuno Carvalhais & Prabhat
Nature volume 566, pages195–204 (2019)

QT:[[”
Figure 3 presents a system-modelling view that seeks to integrate machine learning into a system model. As an alternative perspective, system knowledge can be integrated into a machine learning frame- work. This may include design of the network architecture36,79, physical constraints in the cost function for optimization58, or expansion of the training dataset for undersampled domains (that is, physically based data augmentation)80.
…
Surrogate modelling or emulation
See Fig. 3 (circle 5). Emulation of the full (or specific parts of) a physical model can be useful for computational efficiency and tractability rea- sons. Machine learning emulators, once trained, can achieve simulations orders of magnitude faster than the original physical model without sacrificing much accuracy. This allows for fast sensitivity analysis, model parameter calibration, and derivation of confidence intervals for the estimates.
…
(2) Replacing a ‘physical’ sub-model with a machine learning model
See Fig. 3 (circle 2). If formulations of a submodel are of semi-empirical nature, where the functional form has little theoretical basis (for example, biological processes), this submodel can be replaced by a machine learning model if a sufficient number of observations are available. This leads to a hybrid model, which combines the strengths of physical modelling (theoretical foundations, interpretable compartments) and machine learning (data-adaptiveness).
…
Integration with physical modelling
Historically, physical modelling and machine learning have often been treated as two different fields with very different scientific paradigms (theory-driven versus data-driven). Yet, in fact these approaches are complementary, with physical approaches in principle being directly interpretable and offering the potential of extrapolation beyond observed conditions, whereas data-driven approaches are highly flexible in adapting to data and are amenable to finding unexpected patterns (surprises).
…
A success story in the geosciences is weather
prediction, which has greatly improved through the integration of better theory, increased computational power, and established observational systems, which allow for the assimilation of large amounts of data into the modelling system2
. Nevertheless, we can accurately predict the evolution
of the weather on a timescale of days, not months.
“]]

# REFs that I liked
ref 80

ref 57
Karpatne, A. et al. Theory-guided data science: a new paradigm for scientific discovery from data. IEEE Trans. Knowl. Data Eng. 29, 2318–2331 (2017).

# some key BULLETS

• Complementarity of physical & ML approaches
–“Physical approaches in principle being directly interpretable and offering the potential of extrapolation beyond observed conditions, whereas data-driven approaches are highly flexible in adapting to data”

• Hybrid #1: Physical knowledge can be integrated into ML framework –Network architecture
–Physical constraints in the cost function
–Expansion of the training dataset for undersampled domains (ie physically based data augmentation)

• Hybrid #2: ML into physical – eg Emulation of specific parts of a physical for computational efficiency

https://www.nature.com/news/2010/100715/full/news.2010.356.html

QT:[[”

It should be possible to scan the genome for sequences encoding peptides shorter than 100 amino acids, says Mark Gerstein, a computational biologist at Yale University in New Haven, Connecticut, but sorting through the many ‘hits’ to determine which are functional is likely to be much more difficult.

Meanwhile, Gerstein notes that the polished rice peptides could also have implications for how we view pseudogenes, which have long been thought to be defunct relics of protein-coding genes. Pseudogenes often contain many signals that would stop protein synthesis and, as a result, could only encode short amino-acid chains. “Maybe this would provide a new way for pseudogenes to have some sort of function,” he says.
“]]

http://causality.cs.ucla.edu/blog/index.php/2016/06/28/on-the-classification-and-subsumption-of-causal-models/

QT:[[”
The taxonomy that has helped me immensely is the three-level hierarchy described in chapter 1 of my book Causality: 1. association, 2. intervention, and 3 counterfactuals. It is a useful hierarchy because it has an objective criterion for the classification: You cannot answer questions at level i unless you have assumptions from level i or higher.
“]]

https://www.nytimes.com/2019/01/17/magazine/ancient-dna-paleogenomics.html

has an interesting discussion of review process at Nature

QT:[[”
It has not gone unnoticed that the stunning, magisterial sweep of genetic revisionism, on the one hand, and a genetic emphasis on radical prehistoric migrations, on the other, bear more than a little in common. Some anthropologists
and archaeologists accept this analogy with gallows humor. One told me that I should
model this article after the format of the standard Nature paper: “Ancient DNA Reveals Massive Population Turnovers in the Humanities,” she suggested as a title,
and proposed this as an abstract: “The aristocratic lab scientists arrived with their
superior technology and displaced the pre-existing researchers and their primitive
truth-implements and overcomplicated belief systems.
…
Serious challenges to its soundness were laid out during
Nature’s peer-review process. And yet, in a highly unusual move, the paper was accepted over the steadfast objections of two of the three peer reviewers on its anonymous panel. Confidential documents made available to me reveal deep concerns with the paper’s methods and its conclusions.
“]]

QT:[[”
“The authors describe and validate a citation-based index of ‘disruptiveness’ that has previously been proposed for patents6. The intuition behind the index is straightforward: when the papers that cite a given article also reference a substantial proportion of that article’s references, then the article can be seen as consolidating its scientific domain. When the converse is true — that is, when future citations to the article do not also acknowledge the article’s own intellectual forebears — the article can be seen as disrupting its domain.

The disruptiveness index reflects a characteristic of the article’s underlying content that is clearly distinguishable from impact as conventionally captured by overall citation counts. For instance, the index finds that papers that directly contribute to Nobel prizes tend to exhibit high levels of disruptiveness, whereas, at the other extreme, review articles tend to consolidate their fields.”
“]]

http://www.nature.com/articles/d41586-019-00350-3

https://apple.stackexchange.com/questions/76132/how-do-i-find-out-what-computers-i-have-authorized

QT:

same in ’17

https://www.howtogeek.com/187607/what-you-need-to-know-about-deauthorizing-itunes/

“What is important is seldom urgent and what is urgent is seldom important.” — Dwight D. Eisenhower

https://medium.com/the-mission/the-eisenhower-method-for-taking-action-how-to-distinguish-between-urgent-and-important-tasks-895339a13dea

“Is Email Making Professors Stupid?” is the Q posed by
https://www.Chronicle.com/interactives/is-email-making-professors-stupid . My A: YES. The article has a nice description of the problem with 24/7 connectivity: how the urgent but unimportant crowds out the important but non-urgent

QT:(((”
“Knuth does provide his mailing address at Stanford, and he asks that people send an old-fashioned letter if they need to contact him. His administrative assistant gathers these letters and presents them to Knuth in batches, getting urgent correspondence to him quickly, and putting everything else into a “buffer” that he reviews, on average, “one day every three months.”

Knuth’s approach to email prioritizes the long-term value of uninterrupted concentration over the short-term convenience of accessibility. Objectively speaking, this tradeoff makes sense, but it’s so foreign to most tenured and tenure-track professors that it can seem ludicrous — more parody than pragmatism. This is because in the modern academic environment professors act more like middle managers than monastics. A major factor driving this reality is the digital communication Knuth so carefully avoids. Faculty life now means contending with an unending stream of electronic missives, many of which come with an expectation of rapid reply.”
“)))

QT:((”
“One reason is the rebound effect. Insulate buildings better, and people will wear fewer layers rather than turn the heat down. The way energy-efficiency schemes are structured does not help, argues Richard Twinn of the UK Green Building Council, a think-tank. The schemes only finance a single type of upgrade at a time, such as loft insulation. A whole-house retrofit, in contrast, could have added digital
thermostats to ensure that greater efficiency was converted into lower bills rather than higher temperatures.”
“))

Efforts to make buildings greener are not working
https://www.economist.com/international/2019/01/05/efforts-to-make-buildings-greener-are-not-working

Good quotes:
QT:((
Hoover noted, “When we are sick, we want an uncommon doctor; when we have a construction job to do, we want an uncommon engineer; and when we are at war, we want an uncommon general. It is only when we get into politics that we are satisfied with the common man.”
…
There’s just too much to do. Instead, presidents should follow Calvin Coolidge’s model. “Perhaps one of the most important accomplishments of my administration has been minding my own business,” he said. …
Can one person handle all this? In 1955, former President Herbert Hoover completed a review—his second—of executive-branch efficiency and suggested the addition of an administrative vice president to help the overloaded president. (The existing vice president was apparently already too busy.) Hoover’s report was issued a few months before President Eisenhower had his first heart attack. It was the fifth heart attack or stroke to hit a current or former president since the Wilson administration ended, in 1921. This caused the columnist Walter Lippmann to wonder whether the job was too much for one man to bear. Addressing the “intolerable strain” on the president, Lippmann wrote, “The load has become so enormously greater … because of the wars of this century, because of the huge growth of the American population, of the American economy, and of American responsibilities.”
“))

https://www.theatlantic.com/magazine/archive/2018/05/a-broken-office/556883/