Fully Homomorphic Encryption (FHE)
HEIR: Homomorphic Encryption Intermediate Representation
Fully Homomorphic Encryption (FHE) is an emerging cryptographic technique that allows developers to perform computations on encrypted data. This represents a paradigm shift in how data processing and data privacy relate to each other.
Previously, if an application had to perform some computation on data that was encrypted, this application would necessarily need to decrypt the data first, perform the desired computations on the clear data, and then re-encrypt the data. FHE, on the other hand, simply removes the need for this decryption-encryption steps by the application, all at once.
In practice, for an application that needs to perform some computation F on data that is encrypted, the FHE scheme would provide some alternative computation F' which when applied directly over the encrypted data will result in the encryption of the application of F over the data in the clear. More formally: F(unencrypted_data) = Decrypt(F'(encrypted_data)).
HEIR: Homomorphic Encryption Intermediate Representation
Forwarded from HN Best Comments
Re: Cops say criminals use a Google Pixel with GrapheneOS – I say that's freedom
GrapheneOS says
"European authoritarians and their enablers in the media are misrepresenting GrapheneOS and even Pixel phones as if they're something for criminals. GrapheneOS is opposed to the mass surveillance police state these people want to impose on everyone"
https://grapheneos.social/@GrapheneOS/114784469162979608
State employees in their official capacity making inaccurate claims to media about GrapheneOS to smear it as being for criminals and as the users as largely being criminals is a state sponsored attack on the GrapheneOS project.
https://grapheneos.social/@GrapheneOS/114813613250805804
patchtopic, 2 days ago
GrapheneOS says
"European authoritarians and their enablers in the media are misrepresenting GrapheneOS and even Pixel phones as if they're something for criminals. GrapheneOS is opposed to the mass surveillance police state these people want to impose on everyone"
https://grapheneos.social/@GrapheneOS/114784469162979608
State employees in their official capacity making inaccurate claims to media about GrapheneOS to smear it as being for criminals and as the users as largely being criminals is a state sponsored attack on the GrapheneOS project.
https://grapheneos.social/@GrapheneOS/114813613250805804
patchtopic, 2 days ago
GrapheneOS Mastodon
GrapheneOS (@GrapheneOS@grapheneos.social)
European authoritarians and their enablers in the media are misrepresenting GrapheneOS and even Pixel phones as if they're something for criminals. GrapheneOS is opposed to the mass surveillance police state these people want to impose on everyone.
http…
http…
Do It by Code
Telegram has added "Age verification" feature (works for UK only)
Whether the user is required to do age verification or not, is fetched from Telegram servers (via AppConfig)
(screenshot is related to recent changes done to tdesktop)
(screenshot is related to recent changes done to tdesktop)
This media is not supported in your browser
VIEW IN TELEGRAM
Telegram's @botfather can now be used with its miniapp.
their miniapp URL is: https://webappinternal.telegram.org/botfather
their miniapp URL is: https://webappinternal.telegram.org/botfather
Do It by Code
Telegram's @botfather can now be used with its miniapp. their miniapp URL is: https://webappinternal.telegram.org/botfather
Here is a curl request the bot sends for checking a username's availability:
curl 'https://webappinternal.telegram.org/botfather/api?hash=THE_HASH' \
-H 'accept: application/json, text/javascript, */*; q=0.01' \
-H 'accept-language: en-US,en;q=0.9' \
-H 'cache-control: no-cache' \
-H 'content-type: application/x-www-form-urlencoded; charset=UTF-8' \
-b 'stel_bot_father_token=MY_TOKEN; stel_ssid=MY_SSID; stel_dt=-210' \
-H 'origin: https://webappinternal.telegram.org' \
-H 'pragma: no-cache' \
-H 'priority: u=1, i' \
-H 'referer: https://webappinternal.telegram.org/botfather/create' \
-H 'sec-ch-ua: "Chromium";v="138", "Microsoft Edge";v="138", "Microsoft Edge WebView2";v="138", "Not)A;Brand";v="8"' \
-H 'sec-ch-ua-mobile: ?0' \
-H 'sec-ch-ua-platform: "Windows"' \
-H 'sec-fetch-dest: empty' \
-H 'sec-fetch-mode: cors' \
-H 'sec-fetch-site: same-origin' \
-H 'user-agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36 Edg/138.0.0.0' \
-H 'x-requested-with: XMLHttpRequest' \
--data-raw 'username=hellobot&method=checkBotUsername'
Forwarded from HN Best Comments
Re: Google's shortened goo.gl links will stop working next month
About 60k academic citations about to die - https://scholar.google.com/scholar?start=90&q=%22https://goo...
Countless books with irrevocably broken references - https://www.google.com/search?q=%22://goo.gl%22&sca_upv=1&sc...
And for what? The cost of keeping a few TB online and a little bit of CPU power?
An absolute act of cultural vandalism.
edent, 6 hours ago
About 60k academic citations about to die - https://scholar.google.com/scholar?start=90&q=%22https://goo...
Countless books with irrevocably broken references - https://www.google.com/search?q=%22://goo.gl%22&sca_upv=1&sc...
And for what? The cost of keeping a few TB online and a little bit of CPU power?
An absolute act of cultural vandalism.
edent, 6 hours ago
Data-snooping bias
Data snooping bias, also known as data dredging or data fishing, occurs when patterns in a dataset are discovered through repeated analysis, and these patterns are then mistakenly taken as real relationships that can be used for prediction or generalization. This bias arises because, by repeatedly analyzing the same data, one is more likely to find spurious correlations or patterns purely by chance, especially if the analysis is not guided by a pre-defined hypothesis.
Examples:
- Financial Markets: Data snooping bias is a common problem in financial analysis, where traders might analyze historical data to find profitable trading strategies, but these strategies may not work in the future (the chances of happening this increases when you have a more complex strategy with more (unrelated) parameters!)
- Machine Learning: In machine learning, data snooping can lead to "overfitting", where a model performs well on the training data but poorly on new, unseen data.
- Scientific Research: Data snooping can lead to false discoveries in scientific research, especially in fields with large datasets and multiple potential analyses.
Data snooping bias, also known as data dredging or data fishing, occurs when patterns in a dataset are discovered through repeated analysis, and these patterns are then mistakenly taken as real relationships that can be used for prediction or generalization. This bias arises because, by repeatedly analyzing the same data, one is more likely to find spurious correlations or patterns purely by chance, especially if the analysis is not guided by a pre-defined hypothesis.
Examples:
- Financial Markets: Data snooping bias is a common problem in financial analysis, where traders might analyze historical data to find profitable trading strategies, but these strategies may not work in the future (the chances of happening this increases when you have a more complex strategy with more (unrelated) parameters!)
- Machine Learning: In machine learning, data snooping can lead to "overfitting", where a model performs well on the training data but poorly on new, unseen data.
- Scientific Research: Data snooping can lead to false discoveries in scientific research, especially in fields with large datasets and multiple potential analyses.
🤓1
Do It by Code
Data-snooping bias Data snooping bias, also known as data dredging or data fishing, occurs when patterns in a dataset are discovered through repeated analysis, and these patterns are then mistakenly taken as real relationships that can be used for prediction…
Is it possible to solve data-snooping bias?
sure it is, in my opinion, this is the best way you can prevent/solve it:
Let's say you have 10k data units (I'm just giving this as an example, the "unit" here could be anything depending on your situation):
instead of training your model on all these units, split then:
- consider 5k units for training data
- and another 5k units for testing data
see how does your model perform on the other half. If it's not performing well....well, it's a sign that you might have data-snooping bias! maybe your parameters are way too strict or complex? (that may only work on the first half and won't ever be repeated in the future?)
sure it is, in my opinion, this is the best way you can prevent/solve it:
Let's say you have 10k data units (I'm just giving this as an example, the "unit" here could be anything depending on your situation):
instead of training your model on all these units, split then:
- consider 5k units for training data
- and another 5k units for testing data
see how does your model perform on the other half. If it's not performing well....well, it's a sign that you might have data-snooping bias! maybe your parameters are way too strict or complex? (that may only work on the first half and won't ever be repeated in the future?)
The PlayCanvas Game engine, a game engine for 3D WebGPU and WebGL games is now fully open source (its web-editor wasn't open-source previously)
- Repository
- youtube introduction
#game_engine #js #ts
- Repository
- youtube introduction
#game_engine #js #ts
Spotify has implemented age verification by passport, mirroring Telegram's approach. This new rule, now active in the UK, requires users to upload a document photo and undergo facial scanning to avoid permanent account deletion. The move is a response to the "On Safe Internet" law, which introduced various restrictions.