Yutaro Shimizu

5 followers · 13 following · 914 views

on the atlas — 42

highlights — 102

cell c hemistry approaches takes inspiration from reaction- diffusion models ( Turing, 1952 ) and gene regulation networks ( Davidson, 2010 ; Cussat-Blanc et al., 2019 ). In these models, genes produce proteins that either have a direct phenotypic effect or regulate the expres- sion of other genes (
Evolving the Behavior of Machines: From Micro to Macroevolution
e developmental encodings (also called indirect encodings ), which are compressed representations of networks that make it possible to create modular and regular phenotypes from shorter descriptions
Evolving the Behavior of Machines: From Micro to Macroevolution
s into structural units (modules) that can be reused, with or without variation
Evolving the Behavior of Machines: From Micro to Macroevolution
This feat is possible because the relationship between genes and biological networks is not one-to-one; instead, genes are developed to the phenotype through a com- plex genotype to phenotype map ( Ahnert, 2017 ), which is still little understoo
Evolving the Behavior of Machines: From Micro to Macroevolution
only about 25,000 genes are enough to describe the trillions of cells that make up a human ( Southan, 2004 )
Evolving the Behavior of Machines: From Micro to Macroevolution
e perfect topology for a neural network is, howeve r, rarely known. The problem might, for example, require recurrent connections to take time into account, or it might need more hidden neurons, or there might exist solutions with order of magnitudes fewer connections. A
Evolving the Behavior of Machines: From Micro to Macroevolution
The results show that neuroevolution can reach similar scores as state-of-the-art deep reinforcement learning algorithms. In addition, although neuroevolution requires more computation, it runs faster (in term of wall-clock time) when many computer cores are available because it is straightforward to compute the fitness values in parallel
Evolving the Behavior of Machines: From Micro to Macroevolution
s. In addition, neural networks typically distribute their ‘‘knowledge’’ in all the weights, which means that crossing two parents is unlikely to produce a neural network that combine their functionality. As a consequence, most authors choose to not use the crossover operators and only use random mutations
Evolving the Behavior of Machines: From Micro to Macroevolution
‘How can an artificial genotype describe a neural network that could be as complex as a brain?’’—and the stepping stones problem —‘‘what intermediate s teps lead to an arti- fact as sophisticated as a brain
Evolving the Behavior of Machines: From Micro to Macroevolution
s neuroevolution approach considers the value of the policy without attempting to identify the most useful state-action pairs. Evolutionary methods therefore discard potentially useful information contained in the sequence of decisions and states followed by the agen
Evolving the Behavior of Machines: From Micro to Macroevolution
‘‘brain’’ of a machine is often viewed as a reinforcement learning problem ( Sutton and Barto, 2018 ), in which the objective is to learn a policy that allows the agent to maximize its expected reward by exploring its environmen
Evolving the Behavior of Machines: From Micro to Macroevolution
y, evolutionary algorithms can be viewed as a model of natural evolution: all individuals compete against each other, the fittest re produce and least fit disappear; but repeated appli- cation of this simple rule of natural selection leads these algorithms to converge on a single genome, that of the fittest individual, rather than the diversity of species that we observe in nature. Such convergence implies that traditional evolutionary algorithms mo re closely resemble a microevolutionary pro
Evolving the Behavior of Machines: From Micro to Macroevolution
currently considered as a mathematical optimization algorithm, that is, as an algorithm that finds the maximum of a function
Evolving the Behavior of Machines: From Micro to Macroevolution
the wonder of evolution is less in the perfect optimization of each spe- cies than in the creativity of such a simple iterative process, that is, in the diversity of species. This modern view of artificial evolution is moving the field away from microevolution, following a fitness gradient in a niche, to macroevolution, filling many niches with highly different species. It already opened promising applica- tions, like evolving gait repertoires, video game levels for different tastes, and diverse designs for aerodynamic bike
Evolving the Behavior of Machines: From Micro to Macroevolution
Chess does indeed involve these abilities – in humans. But while possessing these general abilities makes it possible to solve chess (and many more problems), by going from the general to the specific, inversely, there is no clear path from the specific to the general. Chess does not require any of these abilities, and can be solved by taking radical shortcuts that run orthogonal to human cognition
[1911.01547] On the Measure of Intelligence
Applied Chemistry at William Paterson University; however, his enduring passion for drawing led him to earn a Master of Fine Arts from the New York Academy of Art
Alphonso Dunn | About
• Don’t wait for the storm to pass; dance in the rain.
The Technium: 103 Bits of Advice I Wish I Had Known
• Speak confidently as if you are right, but listen carefully as if you are wrong.
The Technium: 103 Bits of Advice I Wish I Had Known
• Your growth as a conscious being is measured by the number of uncomfortable conversations you are willing to have.
The Technium: 103 Bits of Advice I Wish I Had Known
• Whenever there is an argument between two sides, find the third side.
The Technium: 103 Bits of Advice I Wish I Had Known
• When you forgive others, they may not notice, but you will heal. Forgiveness is not something we do for others; it is a gift to ourselves.
The Technium: 103 Bits of Advice I Wish I Had Known
サイエンスが社会にいかに大きなインパクトを与えるか。それにいかに対応すべきか。教えられました。
「私」とNature — 長田重一氏：混沌状態をすっきりさせるような研究が好き | Nature 著者インタビュー | Nature | Nature Portfolio
「考えたくねえ」「受け身でいいだろ」「決められねえ」「正解を教えろ」「リスクを取りたくねえ」が、貧しさの象徴になった世界。これは、一片たりとも、人に優しくない。「自由に生きられる」は、いつの間にか「自由に生きねばならない」に変わっていた。
いつの間にか「好きなことをしていい」時代から、「好きなことをしないと豊かになれない」時代に変わった。 | Books&Apps
そういう行為をして得するのって、炎上を見て楽しんでるギャラリーだけで、自分は得していないので、やめておいたほうがいいかなと思っています。
ヘルシーなジェラシーにしちゃったほうがいいよね、という話｜けんすう
ちなみに「ジェラシーなんて感じないもん！」としてしまうと、それはそれで、なんか心の中にモヤモヤが渦巻いて、相手が失敗したときに「それみたことか！」みたいに噴出することがあるので、いったん、健全なジェラシーということにして排出しておくのがオススメです
ヘルシーなジェラシーにしちゃったほうがいいよね、という話｜けんすう
もし、是正したいという場合は、その場の衝動でやるのではなく、じっくりと腰をすえて、社会のために自分の人生の何割かをそこにつぎ込んで正当に暴いていく、という覚悟が必要です。
ヘルシーなジェラシーにしちゃったほうがいいよね、という話｜けんすう
社会の幸せを考えるより、まず自分が幸せになってからのほうがいい」
ヘルシーなジェラシーにしちゃったほうがいいよね、という話｜けんすう
our morals must catch up with our machinery
weaver-1947b.htm
concern for value or purpose
weaver-1947b.htm
of time and space, across political and cultural boundaries
weaver-1947b.htm
publicized, that the average man has inevitably come to believe that science is just about the most spectacularly successful enterprise man ever launched. The fact is, of course, that this conclusion is largely justified
weaver-1947b.htm
he poet has these virtues also, and often turns them to higher uses
weaver-1947b.htm
its practitioners high standards of personal honesty, open -mindedness, focused vision, and love of the
weaver-1947b.htm
n its physical manifestations; it is to be found in its inner spirit.
weaver-1947b.htm
cience is a way of organizing reproducible knowledge about such problems; of focusing and disciplining imagination; of weighing evidence; of deciding what is relevant and what is not; of impartially testing hypotheses; of ruthlessly discarding data that prove to be inaccurate or inadequate; of finding, interpreting, and facing facts, and of making the facts of nature the servants of man
weaver-1947b.htm
predominant factors are subject to the basic laws of logic, and are for the most part
weaver-1947b.htm
those scientists for whom intellectual freedom is necessarily a private affair.
weaver-1947b.htm
it cannot be assumed that this will be the exclusive pattern for future scientific work
weaver-1947b.htm
g applied to the biological, the medical, and even the social sciences. The results are as yet scattered, but they are highly promising
weaver-1947b.htm
Is there any promise on the horizon that this new advance can really be accomplished
weaver-1947b.htm
hey had only made the brave beginnings of quantitative theories, and hardly even begun detailed explanations of the physical and chemical mechanisms underlying or making up biological events
weaver-1947b.htm
several dozen quantities
weaver-1947b.htm
are seldom those in which one can rigidly maintain constant all but two variables.
weaver-1947b.htm
How can we appreciate what science really is and, equally important, what science is not ?
weaver-1947b.htm
the character of mathematics plays a role in our story, along with Nature consonant and conformable to herself.
4: Nature Conformable to Herself — SFI Press
such mathematical structures are comparatively rare, so that it is possible to find and identify something like the exceptional group E8 as an object of interest simply because structures with its remarkable properties are not thick on the ground
4: Nature Conformable to Herself — SFI Press
he study of complex systems astonishing power laws extending over many orders of magnit
4: Nature Conformable to Herself — SFI Press
renormalizable approximate theories
4: Nature Conformable to Herself — SFI Press
Ultimately, then, we can argue that it is the self-similarity of the structure of fundamental physical law that dictates the continuing usefulness of mathematics.
4: Nature Conformable to Herself — SFI Press
old mathematics may still be useful at the next layer
4: Nature Conformable to Herself — SFI Press