Deep learning combines linear functions together, but the way in which they are combined is always nonlinear. If the way in which they were combined were linear, the resulting function would also be linear (a linear combination of linear functions is linear), and you'd have linear regression. Specifically the final layer of a deep neural network is often linear but the nodes in middle layers are often Rectified Linear Units which have a nonlinear "max" function applied to a linear function. Without these "max" functions you'd be able to flatten all the layers into one. Other common nonlinear functions you'll find in a deep neural net are sigmoid functions like tanh. These functions are sometimes called "activation functions".
Have you tried out the ChucK language? I haven't tried Extempore yet, but it looks like the way they handle timing could be similar. ChucK has a very elegant system for synchronizing synthesizers and sequencers where you can write code in a simple imperative style (within a sequencer you can "wait" until the next note needs to be played). The language runtime keeps all of the coroutines running in sync with a global clock. I don't know of other languages that use this style of timing, so I'd be very curious to hear how they compare.
I think it is because many labs in CS departments do very little research involving human subjects (e.g. a machine learning lab or a theory lab), so within those labs there isn't really an expectation that everything goes through IRB. Many CS graduate students likely never have to interact with IRB at all, so they probably don't even know when it is necessary to involve IRB. The rules for what requires IRB involvement are also somewhat open to interpretation. For example, surveys are often exempt depending on what the survey is asking about.
It's very sad Rosenblatt did not live to see the resurgence of neural networks and his perceptron algorithm. The perceptron algorithm isn't exactly what we use to train neural networks today, but it's similar enough in theory and practice that it still feels very fundamental to understanding machine learning.
However, I think the laws for this vary quite a bit across countries, and in many countries the photograph is considered a new copyright work. In general it's pretty frustrating how many legal barriers there are to accessing and reusing old works of art. Thankfully a growing number of museums have made things easy with clear copyright releases (Rijksmuseum, Paris Musées, the MET), but others seem more interested in preserving their ability to sell prints.
This sort of research is maybe less flashy than say using machine learning to automatically generate game assets from photos, but I think this sort of computer-aided game design is possibly the biggest way machine learning will transform video games. As games are becoming bigger and more complicated, the problem of tuning various gameplay parameters explodes exponentially. And, this sort of tuning can have a huge effect on player retention and overall game quality.
In this research the machine learning is being used to balance the game across different asymmetric strategies (different decks in the card game), but you could imagine using similar techniques for balancing and tuning content for single player games as well. Once you have a reasonable model of the player's behavior, you can do all sorts of automatic tuning like balancing the difficulty of jumps in a platformer, tuning enemy positions in an FPS, etc.
I'm guessing information isn't put in the email because of federal privacy laws around education (FERPA). Email isn't considered a secure communication protocol since it isn't encrypted in transit, so schools can't email grades or other information related to education records. I'm guessing sending all messages through a secure web portal is just an easy way to avoid FERPA liability from the schools perspective.
Dip pens are used quite a bit for artwork still. Until very recently (last 10 years or so), dip pens were very much the standard for inking certain types of art, manga especially. The main advantages of a dip pen are the amount of control you have over the line by varying pressure and angle (like a flexible nib fountain pen) as well as the ability use pigment based inks that will clog most fountain pens. Of course digital tablets have pressure and tilt sensors, but without the tactile feedback of the nib flexing and pushing back against your hand, it can be much harder to control. I'm guessing that drawing tablets will eventually have haptic feedback for this reason. The ability to swap out nibs or use a paint brush with the same ink is also very useful.
Basically, if you wear goggles that distort your vision for long enough, your visual processing adapts and "corrects" for the distortion such that you can function normally. Then if you remove the goggles your brain still tries to correct for the distortion that isn't there (for a while at least), so the world appears distorted without the glasses.
I've found Resolve to be more productive than other video editing programs I've used, primarily because the core operations of navigating through video and cutting clips just feel much more responsive to me. With other editing programs I often feel like the UI is waiting for editing operations to complete in the background, but with Resolve it feels like the UI updates instantly so I can continue working. It sounds like a small change, but it can make a huge difference for productivity since you are no longer waiting for UI confirmation after every action.
Something I see frequently missing from the discussion on roguelikes is the long history of Japanese console roguelikes like the Mystery Dungeon series dating back to 1993. These games have not been very successful outside of Japan but represent some of the earliest successful commercial games inspired by Rogue. They tend to be much closer to traditional "Berlin interpretation" roguelikes (turn-based, grid-based, etc.) than the recent "roguelite" indie games.
To add another item to the list of legal, legitimate uses of YouTube-dl: lots of public domain content is uploaded to YouTube, including a lot of media produced by the US government. For example, The White House has a YouTube channel, and my understanding of US law is that the vast majority of the content uploaded to that channel is public domain (produced by federal employees in the course of their job). Journalists or anyone else wanting to monitor the government would likely find YouTube-dl useful for archiving this public domain content from government channels.
Very cool app--wanted to leave you some feedback after my first test render finished.
Loved the visuals. I can tell that your analysis separates out harmony and percussion as advertised as the video reacts very well to transients in music. This is a common criticism I have of music visualizers--often they don't react to transients quickly enough to communicate the percussion patterns and that's not an issue here.
The whole process creating the video was very easy. Some more control over the images used would be cool, but I'm not sure how that'd work. I do also like that the images are generated and unique for each project. Perhaps if it's possible there could be an advanced feature to upload a pool of images to seed the generation of images. For example, if I wanted to create a video using landscape images of a particular type (mountains, forests, etc), this kind of feature would be very useful.
I'm not sure I'll be a paying customer for this product, as I don't have a commercial use for music visualization at the moment, but I'll be playing around with it some more for sure. I have a big collection of synthesized versions of classical music I created for a project that never materialized, so I can see using this tool to revisit that music and perhaps create videos to share. Thanks for sharing this very cool project!
It could be because of my browser settings / extensions, but I couldn't find any information about the games you can play on the site. It looks like you have to have a paid account to see the list of available games. From a guest account the "Play a Game" page is blank. I would suggest making the games themselves more front and center, as right now the site almost feels like a business-to-business site telling me more about the company itself than the products.
One of the colors that seems to have been lost over the past 20 years or so is Quinacridone Gold, denoted by pigment number PO 49. The pigment is part of the family of quinacridone pigments used everywhere, but reportedly this gold hue variation of the pigment wasn't popular enough to sustain production despite being used by artist paint manufacturers. Paint companies typically buy pigments from a relatively small number of pigment manufacturers. From the perspective of pigment manufacturers, paint for artists is a small market in comparison to other uses like the automotive industry. When the pigment went out of production, a watercolor paint company in Seattle, Daniel Smith, purchased the last batch and had a roughly 17 year monopoly on the color. Of course, artists figured out that the pigment was going away and started buying up the last of the stock, and now it is considered an "extinct pigment," no longer available for purchase in any new paint. There is a reformulated "New Quinacridone Gold" that uses a mixture of two pigments (PO 48 + PY 150), but some artists swear by single pigment colors for their more predictable mixing properties.
I'm completely out of my area of expertise here (I can mostly follow the press release but I can't understand that much of the actual paper outside of the introduction), but your comment about resonant frequencies reminded me of of some recent research on layering graphene at a "magic angle" to produce special fractal moire patterns that have special physical properties. I wonder if that work can be combined with the work here. To a non expert like myself, it seems like there is a lot of really exciting research in this area at the moment.
From my understanding of the article, they aren't necessarily the same but could be related:
Another more speculative possibility is that cosmic strings could come from the tiny vibrating strings of string theory. Some string theory models propose that strings could have grown to colossal proportions during the initial rapid expansion of the cosmos. Differences in the tension of these kinds of cosmic strings and in how string loops break away would create a unique gravitational-wave signature distinguishing them from other kinds.
An interesting aspect to this story is that although there's no real evidence that a Facebook employee was involved, it still seems like a believable explanation to many readers including many commenters here. If a company's customer support is so bad that no one can tell the difference between being hacked and being abused by a rogue employee, does it actually matter what happened? I guess that it matters to the original poster, and I do hope that they do get their photos / account back, but in either case the message the message I'm taking away from the story same: your Facebook account could disappear tomorrow and you'd have no recourse.
The history behind the term is pretty interesting to me: https://en.wikipedia.org/wiki/Dynamic_programming#History The original research on the subject was published in engineering / optimization as opposed to computer science, and the term "programming" was used differently there ("linear programming" and "quadratic programming" for example). Now we think of it not as an optimization problem but as an algorithm design technique, but the name still stuck around.
What goes on at the orientation? I'm guessing from your screen name you have some first hand experience. This blog post is my first time hearing about the Recurse Center so I don't know how it differs from other similar programs.
Not sure how all the services work, but I believe the drivers are given delivery job offers where they see information about the delivery location and payout estimate, and they're able to accept / reject jobs based on this info. Therefore it's not necessarily that there weren't drivers available just that the drivers chose different jobs. It might be that you're in an area that isn't particularly profitable for deliveries (homes or restaurants too spread out). I believe you can also tip more / tip upfront to make it more likely that drivers accept the job. Although these services present tipping as an optional gratuity, if you don't tip up front it could lead to your job sitting in the queue forever since it's not as appealing to drivers as other available delivery jobs with included tips.
Thank you for the reference! Your explanation makes sense to me. I'm curious if students that struggle with learning decimal expansions and limits are just subconsciously rejecting the non-constructive foundations of the math they're learning. Probably not, but it's fascinating regardless.
I wonder if this is related to "intuitionist" math. This is an alternative formulation of math which doesn't have the law of excluded middle, recently discussed on Hacker News relating to this physics research: https://www.quantamagazine.org/does-time-really-flow-new-clu...
"Online learning" is actually a pretty established term in theoretical computer science and specifically learning theory. I don't think that it's meant to be misleading at all but rather to connect the work to the extensive previous work on this subject. "Online" is used in general in theoretical computer science to describe problems where data arrives in a stream and the algorithm has to make decisions as it observes the data (e.g. before observing all the data). There are "online" versions of just about every problem in theoretical computer science. For example, there are "online traveling salesman" and "online facility location" problems both of which have published research. "Online" obviously has a different meaning outside of computer science, but it conveys a very specific meaning in the math / computer theory context. I'm not sure when this term first appeared in research, but this line of work is often traced back to the 50s with the perceptron algorithm, David Blackwell, and James Hannan.
I would also add to this that in my opinion often the best games (from a design perspective) are the ones that lean into this sort of freedom. This line of thinking has very much influenced game design trends, at least within certain genres. For example, simulation and survival games usually try to expose the game's systems to the player in a way that lets the player be creative in the game (Minecraft, Prison Architect, RimWorld, Factorio, etc). Even in games that aren't based on simulation or user generated content, designing the game's systems in this way gives the level creators freedom to do more without having to write code specific for every level.