317 Matching Annotations
  1. Sep 2023
  2. Aug 2023
    1. Title: Delays, Detours, and Forks in the Road: Latent State Models of Training Dynamics Authors: Michael Y. Hu1 Angelica Chen1 Naomi Saphra1 Kyunghyun Cho Note: This paper seems cool, using older interpretable machine learning models, graphical models to understand what is going on inside a deep neural network

      Link: https://arxiv.org/pdf/2308.09543.pdf

  3. Jul 2023
    1. Figure 10.7 The flow velocltles in the optic array reflectedfrom the surface of the earth. This is the flow pattern obtainedwith locomotion parallel to the earth in the direction of the pole atthe top of the graph. The vectors are plotted in angular coordinates.This is a view from above, whereas Figure 9.3 was a view from theside.

      Visual of the flow velocities in the optic array.

    2. Figure 10.5 The effective array at a stationary convergence point(Stage 5). The solid lines represent the sample of the total optic arraythat is admitted to a human eye in a given posture. The dashed linesrepresent the remainder of the array, which is available for stimulationbut not effective at this moment.

      Visual example of an "effective array" along with the broader "optic array".

    1. A paper recommended by Pablo B. which is about the implementation of a software for iterative visualizations. Looks very promising

    2. The resulting system effec-tively models diagram genera-tion as a compilation process,where the compilation targetis a constrained optimizationproblem rather than (say) a bi-nary executable or a static im-age.

      A neat idea of using compilation to target [[constrained optimization problem]] instead of a binary executable or static image.

    1. A visual depiction of the transition from effortful to effortless meditation.

      Interesting that the [[default mode network]] and [[frontoparietal network]] connectivity goes down (from naïve to developing meditator) and then goes up much higher (from developing meditator to proficient meditator). Seems like the DMN is suppressed to allow for more CEN/FN activation during a developing meditator's practice. Seems like the eventual transition of allowing co-activation of the DMN and the CEN/FN is associated with the "effortless" described in [[nondual awareness]] [[meditation]] like what [[Diana Winston]] calls [[natural awareness]] in the series [[The Spectrum of Awareness]] or what [[Loch Kelly]] teaches in his series [[Effortless Mindfulness]] (both of which are on the [[Waking Up app]]).

    1. This was a tool mentioned by [[Grant Sanderson]] in the description of a video on [[divergence]] and [[curl]]. Helpful tool for visualizing [[vector fields]].

    1. https://erinflotodesigns.com/collections/metal-stencils/products/bullet-journal-monthly-circle-tracker-stencil-habit-tracker-stencil-unique-design-stencil-cleaning-tracker-stencil

      Erin Floto has a metal stencil for a chronodex circular design for use in bullet journals. It's a form of circular calendar with the inner circle containing space for daily, bi-weekly, weekly, monthly and longer time horizons with succeeding rings of the circle containing space for data related to the inner categories. Some of the exterior rings also include numbered squares representing days of the month or week on which a task should be done or for which a habit on an interior part of the circle might be tracked.

      The chronodex, a portmanteau of chrono (time) and index, idea is fairly simple, but can be quite complex. For actual use, one may need to be able to spin the visualization around to read and understand it.

      Other stencils with habit trackers, etc: https://erinflotodesigns.com/collections/metal-stencils

  4. Jun 2023
    1. Found this webpage for a 3D brain model when someone (maybe frymatic?) mentioned a region of the brain I was having trouble imagining.

  5. May 2023
    1. https://www.napkin.one/

      Yet another collection app that belies the work of taking, making, and connecting notes.

      Looks pretty and makes a promise, but how does it actually deliver? How much work and curation is involved? What are the outputs at the other end?

  6. Apr 2023
    1. Now we are getting somewhere. At this point, we also see that the dimensions of W and b for each layer are specified by the dimensions of the inputs and the number of nodes in each layer. Let’s clean up the above diagram by not labeling every w and b value individually.
  7. Mar 2023
    1. There are two main reasons to use logarithmic scales in charts and graphs.
      • respond to skewness towards large values / outliers by spreading out the data.
      • show multiplicative factors rather than additive (ex: b is twice that of a).

        The data values are spread out better with the logarithmic scale. This is what I mean by responding to skewness of large values.

      In Figure 2 the difference is multiplicative. Since 27 = 26 times 2, we see that the revenues for Ford Motor are about double those for Boeing. This is what I mean by saying that we use logarithmic scales to show multiplicative factors

    2. One reason for choosing a dot plot rather than a bar chart is that it is less cluttered. We will be learning other benefits of dot plots in this and future posts.
      • Length of bar/line has no meaning in a log-scale

        A dot plot is judged by its position along an axis; in this case, the horizontal or x axis. A bar chart is judged by the length of the bar. I don’t like using lengths with logarithmic scales. That is a second reason that I prefer dot plots over bar charts for these data.

    1. a Structure Note can make use of a TOC form, a normal table, a mind map, a flow diagram, a straight list, or even a picture.

      Structure notes can take a variety of forms including lists, diagrams, mind maps, tables, and tables of contents.

  8. Jan 2023
    1. Data Viz with Python and RLearn to Make Plots in Python and R

      data viz with python and R

  9. Nov 2022
    1. An independent initiative made by Owen Cornec who has also made many other beautiful data visualizations. Wikiverse vividly captures the fact that Wikipedia is a an awe-inspiring universe to explore.

    1. The only diagram or image in The Origin of Species, a tree depicting divergence (source)

      Darwin's On the Origin of Species only contains one diagram, a branching tree diagram which shows divergence of species.

    2. he was working on the same theme with Stefanie Posavec. They completed their piece some time later, depicting the changes as lovely branching trees — a kind of homage to Darwin’s lone diagram in the book.

      Greg McInerny of Microsoft Research and Stefanie Posavek created a version of Darwin's On the Origin of Species that displayed variations between the editions as a branching tree diagram, a nod to the only diagram which appeared in Darwin's original work. .

  10. Oct 2022
    1. https://www.denizcemonduygu.com/philo/browse/

      History of Philosophy: Summarized & Visualized

      This could be thought of as a form of digital, single-project zettelkasten dedicated to philosophy. It's got people, sources, and ideas which are cross linked in a Luhmann-sense (without numbering) though not in a topical index-sense. Interestingly it has not only a spatial interface and shows spatial relationships between people and ideas over time using a timeline, but it also indicates—using colored links—the ideas of disagreement/contrast/refutation and agreement/similarity/expansion.

      What other (digital) tools of thought provide these sorts of visualization affordances?

  11. cosma.graphlab.fr cosma.graphlab.fr
    1. https://cosma.graphlab.fr/<br /> https://cosma.graphlab.fr/en/

      When did this come out?

      Appears to be a visualization tool for knowledge work. They recommend it for use with Zettlr, but it looks like it would work with other text based tools. Point it at markdown files to create graphs apparently.

      This looks like the sort of standards based tool that would allow greater flexibility when using various data stores that we talk about in Friends of the Link.

      <small><cite class='h-cite via'> <span class='p-author h-card'>Arthur Perret </span> in And you, what are you doing? (<time class='dt-published'>08/31/2022 02:40:03</time>)</cite></small>


    1. For physical note taking on index cards or visualizations provided by computer generated graphs, one can physically view a mass of notes and have a general feeling if there is a large enough corpus to begin writing an essay, chapter, or book or if one needs to do additional research on a topic, or perhaps pick a different topic on which to focus.

      (parts suggested by p7, though broadly obvious)

    1. here are several ways I havefound useful to invite the sociological imagination:

      C. Wright Mills delineates a rough definition of "sociological imagination" which could be thought of as a framework within tools for thought: 1. Combinatorial creativity<br /> 2. Diffuse thinking, flâneur<br /> 3. Changing perspective (how would x see this?) Writing dialogues is a useful method to accomplish this. (He doesn't state it, but acting as a devil's advocate is a useful technique here as well.)<br /> 4. Collecting and lay out all the multiple viewpoints and arguments on a topic. (This might presume the method of devil's advocate I mentioned above 😀)<br /> 5. Play and exploration with words and terms<br /> 6. Watching levels of generality and breaking things down into smaller constituent parts or building blocks. (This also might benefit of abstracting ideas from one space to another.)<br /> 7. Categorization or casting ideas into types 8. Cross-tabulating and creation of charts, tables, and diagrams or other visualizations 9. Comparative cases and examples - finding examples of an idea in other contexts and time settings for comparison and contrast 10. Extreme types and opposites (or polar types) - coming up with the most extreme examples of comparative cases or opposites of one's idea. (cross reference: Compass Points https://hypothes.is/a/Di4hzvftEeyY9EOsxaOg7w and thinking routines). This includes creating dimensions of study on an object - what axes define it? What indices can one find data or statistics on? 11. Create historical depth - examples may be limited in number, so what might exist in the historical record to provide depth.

  12. Sep 2022
  13. www.mariakozhevnikov.com www.mariakozhevnikov.com
    1. As arousal-based meditative practices are associated with a different autonomic state to mindfulness-related practices (PNS withdrawal vs. PNS dominance), they are likely to recruit a different type of alerting attention (selective or focused vs. sustained) and attentional control (modulated by arousal vs. regulated by monitoring continuous thought processes). In particular, my focus is on Vajrayana (Tantric Buddhism) practices, during which a sequence of generation (self-visualization as a deity - Yidam) or completion with sign (inner heat -Tummo) stages necessarily precedes non-dual awareness (NDA) Tantric Mahamudra. I conduct my studies with experienced monks and nuns in Eastern Tibet, Bhutan, and Nepal. I studied these practices using cognitive (visual working memory and attention) assessments as well as electrocardiographic (EKG) and (EEG) measures.

      !- for : visualization practice - could this research benefit practitioners who have a hard time visualizing ?

  14. Aug 2022
    1. http://cluster.cis.drexel.edu/~cchen/talks/2011/ICSTI_Chen.pdf

      The Nature of Creativity: Mechanism, Measurement, and Analysis<br /> Chaomei Chen, Ph.D.<br /> Editor in Chief, Information Visualization<br /> College of Information Science and Technology, Drexel University<br /> June 7‐8, 2011

      Randomly ran across while attempting to source Randall Collins quote from https://hypothes.is/a/8e9hThZ4Ee2hWAcV1j5B9w

  15. Jul 2022
  16. Jun 2022
    1. 80% of data analysis is spent on the process of cleaning and preparing the data

      Imagine having unnecessary and wrong data in your document, you would most likely have to experience the concept of time demarcation -- the reluctance in going through every single row and column to eliminate these "garbage data". Clearly, owning all kinds of data without organizing them feels like stuffing your closet with clothes that you should have donated 5 years ago. It is a time-consuming and soul-destroying process for us. Luckily, in R, we have something in R called "tidyverse" package, which I believe the author talks about in the next paragraph, to make life easier for everyone. I personally use dplyr and ggplot2 when I deal with data cleaning, and they are extremely helpful. WIthout these packages' existence, I have no idea when I will be able to reach the final step of data visualization.

  17. May 2022
    1. Recognizing that the CEC hyperthreat operates at micro and macro scales across most forms of human activity and that a whole-of-society approach is required to combat it, the approach to the CEC hyperthreat partly relies on a philosophical pivot. The idea here is that a powerful understanding of the CEC hyperthreat (how it feels, moves, and operates), as well as the larger philosophical and survival-based reasons for hyper-reconfiguration, enables all actors and groups to design their own bespoke solutions. Consequently, the narrative and threat description act as a type of orchestration tool across many agencies. This is like the “shared consciousness” idea in retired U.S. Army general Stanley A. McChrystal’s “team of teams” approach to complexity.7       Such an approach is heavily dependent on exceptional communication of both the CEC hyperthreat and hyper-response pathways, as well as providing an enabling environment in terms of capacity to make decisions, access information and resources. This idea informs Operation Visibility and Knowability (OP VAK), which will be described later.  

      Such an effort will require a supporting worldwide digital ecosystem. In the recent past, major evolutionary transitions (MET) (Robin et al, 2021) of our species have been triggered by radical new information systems such as spoken language, and then inscribed language. Something akin to a Major Competitive Transitions (MCT) may be required to accompany a radical transition to a good anthropocene. (See annotation: https://hyp.is/go?url=https%3A%2F%2Fwww.frontiersin.org%2Farticles%2F10.3389%2Ffevo.2021.711556%2Ffull&group=world)

      If large data is ingested into a public Indyweb, because Indyweb is naturally a graph database, a salience landscape can be constructed of the hyperthreat and data visualized in its multiple dimensions and scales.

      Metaphorically, it can manifest as a hydra with multiple tentacles reach out to multiple scales and dimensions. VR and AR technology can be used to expose the hyperobject and its progression.

      The proper hyperthreat is not climate change alone, although that is the most time sensitive dimension of it, but rather the totality of all blowbacks of human progress...the aggregate of all progress traps that have been allowed to grow, through a myopic prioritization of profit over global wellbeing due to the invisibility of the hyperobject, from molehills into mountains.

  18. Apr 2022
    1. Ourpre-trained network is nearly identical to the “AlexNet”architecture (Krizhevsky et al., 2012), but with local re-ponse normalization layers after pooling layers following(Jia et al., 2014). It was trained with the Caffe frameworkon the ImageNet 2012 dataset (Deng et al., 2009)
    1. Trevor Bedford. (2022, January 10). Given ~680k cases per day, this would in turn suggest 0.8% or 1% of the US being infected with SARS-CoV-2 every day. This would translate to perhaps 5% or 10% of individuals currently infected with SARS-CoV-2 in the US. 15/15 [Tweet]. @trvrb. https://twitter.com/trvrb/status/1480610448563060738

    1. Starting from random noise, we optimize an image to activate a particular neuron (layer mixed4a, unit 11).

      And then we use that image as a kind of variable name to refer to the neuron in a way that more helpful than the the layer number and neuron index within the layer. This explanation is via one of Chris Olah's YouTube videos (https://www.youtube.com/watch?v=gXsKyZ_Y_i8)

  19. Mar 2022
  20. Feb 2022
  21. Jan 2022
    1. refracted ocean light

      I often connect refracted ocean light to summer days at the beach. My body immediately relaxes and that feeling of "nothing to do, nothing to complete" sets in.

    2. haunted house shrieks

      This is such a visceral image. I can hear it and see it.

    1. Zimmerman, M. I., Porter, J. R., Ward, M. D., Singh, S., Vithani, N., Meller, A., Mallimadugula, U. L., Kuhn, C. E., Borowsky, J. H., Wiewiora, R. P., Hurley, M. F. D., Harbison, A. M., Fogarty, C. A., Coffland, J. E., Fadda, E., Voelz, V. A., Chodera, J. D., & Bowman, G. R. (2021). SARS-CoV-2 simulations go exascale to predict dramatic spike opening and cryptic pockets across the proteome. Nature Chemistry, 13(7), 651–659. https://doi.org/10.1038/s41557-021-00707-0

    1. Dr. Cecília Tomori. (2021, December 27). Maryland—Just awful to watch what’s unfolding. Now at 1714 hospitalizations ⬆️ 130 in 24 hrs. 16.5% test positivity. Some counties have acted but no statewide 😷 policy! No measures to slow the spread. Https://coronavirus.maryland.gov https://t.co/C03cSRO2AX [Tweet]. @DrTomori. https://twitter.com/DrTomori/status/1475503877977948166

    1. Ariel Karlinsky. (2022, January 2). Russia at 1.04 MILLION excess deaths since March 2020, which is about 240% higher than their reported COVID-19 deaths. This is 1st place worldwide (for countries with data) in absolute excess mortality, 2nd place on per capita terms and 9th on p-score. #poptwitter #epitwitter https://t.co/aLBRRht3z2 [Tweet]. @ArielKarlinsky. https://twitter.com/ArielKarlinsky/status/1477531141510946818

  22. Dec 2021
    1. Trisha Greenhalgh. (2021, December 27). This is nothing short of scandalous. Unless and until those leading the public health response acknowledge the AIRBORNE nature of the virus and give transmission mitigation advice commensurate with how airborne viruses spread, we will be yo-yoing from wave to wave ad infinitum. [Tweet]. @trishgreenhalgh. https://twitter.com/trishgreenhalgh/status/1475502337594646528

    1. Colin Davis. (2021, December 20). Update for 20th Dec. The trend line still reflects 1.8 day doubling (it’s 1.7 days if we look at just the last week). Today’s number is down, but I wouldn’t read too much into that at this point. Https://t.co/kOCjxhRbop [Tweet]. @ProfColinDavis. https://twitter.com/ProfColinDavis/status/1472969632705392640

    1. Tom Moultrie. (2021, December 17). A 1-figure Gauteng update, bringing in data through Wednesday 15/12 (PCR only; by date of collection). The turn continues. On similar metrics (not shown) ALL northern provinces (NW, GT, MP, LP) seem to have now turned. Https://t.co/6Bh3kZsooK [Tweet]. @tomtom_m. https://twitter.com/tomtom_m/status/1471723711287996416

    1. 100 000+ datapoints). This library solves this by downsampling the signal for the currently selected time window and then plotting the downsampled points.

      Optimization plotting library.

    1. Dave Keating. (2021, December 8). Boris Johnson’s continued pretence that UK is one of the most vaccinated countries in the world, repeated again in press conference just now announcing new restrictions, is getting tiresome. That has not been the case for many many months, despite 🇬🇧🇺🇸 vaccine hoarding early on. Https://t.co/tQt6aXGtNI [Tweet]. @DaveKeating. https://twitter.com/DaveKeating/status/1468655107436802052

    1. Art Poon. (2021, November 28). Our first https://filogeneti.ca/CoVizu update with B.1.1.529. As expected, number of mutations is well over molecular clock prediction (~13 diffs). Relatively low numbers of identical genomes implies large number of unsampled infections. We update every two days from GISAID. https://t.co/m8w2CjL1c0 [Tweet]. @art_poon. https://twitter.com/art_poon/status/1465001066194481162

    1. nference. (2021, November 27). Here is how B.1.1.529 (#Omicron #B11529) compares to Alpha, Beta, Gamma, Delta variants. Omicron has highest novel Spike mutations including striking cluster on the “crown” suggesting significant selection pressure & antigenic distinction from prior strains (Credits: Nference) https://t.co/4oZQbjhbG8 [Tweet]. @_nference. https://twitter.com/_nference/status/1464404770098229250

  23. Nov 2021
    1. Prof. Christina Pagel. (2021, November 24). Meanwhile AY.4.2 (Delta grandchild) continues its very slow path to English dominance. Makes life a bit harder by being a bit more transmissible but luckily doesn’t seem any worse than Delta in any other respect. Https://t.co/kB0V0Z66GT [Tweet]. @chrischirp. https://twitter.com/chrischirp/status/1463508172941967365

    1. The following figure presents a simple functional diagram of the neural network we will use throughout the article. The neural network is a sequence of linear (both convolutional A convolution calculates weighted sums of regions in the input. In neural networks, the learnable weights in convolutional layers are referred to as the kernel. For example Image credit to https://towardsdatascience.com/gentle-dive-into-math-behind-convolutional-neural-networks-79a07dd44cf9. See also Convolution arithmetic. and fully-connected A fully-connected layer computes output neurons as weighted sum of input neurons. In matrix form, it is a matrix that linearly transforms the input vector into the output vector. ), max-pooling, and ReLU First introduced by Nair and Hinton, ReLU calculates f(x)=max(0,x)f(x)=max(0,x)f(x)=max(0,x) for each entry in a vector input. Graphically, it is a hinge at the origin: Image credit to https://pytorch.org/docs/stable/nn.html#relu layers, culminating in a softmax Softmax function calculates S(yi)=eyiΣj=1NeyjS(y_i)=\frac{e^{y_i}}{\Sigma_{j=1}^{N} e^{y_j}}S(yi​)=Σj=1N​eyj​eyi​​ for each entry (yiy_iyi​) in a vector input (yyy). For example, Image credit to https://ljvmiranda921.github.io/notebook/2017/08/13/softmax-and-the-negative-log-likelihood/ layer.

      This is a great visualization of MNIST hidden layers.

    1. COVID-19 Living Evidence. (2021, November 12). As of 12.11.2021, we have indexed 257,633 publications: 18,674 pre-prints 238,959 peer-reviewed publications Pre-prints: BioRxiv, MedRxiv Peer-reviewed: PubMed, EMBASE, PsycINFO https://t.co/ytOhLG90Pi [Tweet]. @evidencelive. https://twitter.com/evidencelive/status/1459163720450519042