A prior-free mechanism (PFM) is a mechanism in which the designer does not have any information on the agents' valuations, not even that they are random variables from some unknown probability distribution. A typical application is a seller who wants to sell some items to potential buyers. The seller wants to price the items in a way that will maximize his profit. The optimal prices depend on the amount that each buyer is willing to pay for each item. The seller does not know these amounts, and cannot even assume that the amounts are drawn from a probability distribution. The seller's goal is to design an auction that will produce a reasonable profit even in worst-case scenarios. PFMs should be contrasted with two other mechanism types: * Bayesian-optimal mechanisms (BOM) assume that the agents' valuations are drawn from a known probability distribution. The mechanism is tailored to the parameters of this distribution (e.g, its median or mean value). * Prior-independent mechanisms (PIM) assume that the agents' valuations are drawn from an unknown probability distribution. They sample from this distribution in order to estimate the distribution parameters. From the point-of-view of the designer, BOM is the easiest, then PIM, then PFM. The approximation guarantees of BOM and PIM are in expectation, while those of PFM are in worst-case. What can we do without a prior? A naive approach is to use statistics: ask the potential buyers what their valuations are and use their replies to calculate an empirical distribution function. Then, apply the methods of Bayesian-optimal mechanism design to the empirical distribution function. The problem with this naive approach is that the buyers may behave strategically. Since the buyers' answers affect the prices that they are going to pay, they may be incentivized to report false valuations in order to push the price down. The challenge in PFMD is to design truthful mechanisms. In truthful mechanisms, the agents cannot affect the prices they pay, so they have no incentive to report untruthfully. Several approaches for designing truthful prior-free mechanisms are described below. (Wikipedia).
(ML 19.2) Existence of Gaussian processes
Statement of the theorem on existence of Gaussian processes, and an explanation of what it is saying.
From playlist Machine Learning
Thermodynamics 4c - Entropy and the Second Law III
We consider in more detail how the fundamental laws of mechanics cannot account for the irreversibility of a system. Yet we find evidence that "special" states are easily transformed into "non-special" states while transforming a non-special state into a special state requires "fine-tuning
From playlist Thermodynamics
Intermittent Planetary Mechanism
This mechanism produces a reciprocating movement, with the forward always longer than the backward. It uses a planetary mechanism with two inputs, the sun and the ring. The output is the arm. The inputs are provided by an intermittent mechanism, with one gear moving two others, one at a ti
From playlist Planetary Mechanisms
Review of Linear Time Invariant Systems
http://AllSignalProcessing.com for more great signal-processing content: ad-free videos, concept/screenshot files, quizzes, MATLAB and data files. Review: systems, linear systems, time invariant systems, impulse response and convolution, linear constant-coefficient difference equations
From playlist Introduction and Background
Definition of conjugate priors, and a couple of examples. For more detailed examples, see the videos on the Beta-Bernoulli model, the Dirichlet-Categorical model, and the posterior distribution of a univariate Gaussian.
From playlist Machine Learning
Thermodynamics 4b - Entropy and the Second Law II
We compare the reversibility of the Carnot cycle to the irreversibility of the Stirling cycle and find that they may be accounted for by the constancy or increase of transferred heat divided by temperature. We then consider how conservation laws, including the fundamental laws of mechanics
From playlist Thermodynamics
This mechanism directly converts the continuous rotary motion of a drive shaft into the intermittent linear motion of a rack. STEP files of this video: http://www.mediafire.com/file/1c0iaa3teed88el/RatchetMechanism6STEP.zip Inventor files: http://www.mediafire.com/file/ujcw5wcp8nabb65/Ratc
From playlist Mechanisms
This mechanism is used in hand powered electric torches to convert oscillatory motion into continuous rotation. STEP files of this video: http://www.mediafire.com/file/qo3kwb01k9pen57/RatchetMechanism10STEP.zip/file Inventor files: http://www.mediafire.com/file/enoia1bi94oc668/RatchetMech
From playlist Mechanisms
Bitcoin Q&A: Watchtowers, Justice Transactions, and Privacy on the Lightning Network
Watchtowers, justice transactions, and privacy, how are these three things balanced on the Lightning Network? How do watchtowers detect fraudulent activity and make justice transactions? What data do you need to give a watchtower? And what are the privacy concerns of using a watchtower ser
From playlist Bitcoin Q&A
Occam's Razor and Statistical Mechanics - Singular Learning Theory Seminar 38
Edmund Lau presents Balasubramanian's paper "Statistical Inference, Occam’s Razor, and Statistical Mechanics on the Space of Probability Distributions". This involves - The principle of Bayesian model selection - Local behaviour of KL divergence / infinitesimal metric / Fisher information
From playlist Singular Learning Theory
Nexus Trimester - Kamalika Chaudhuri (UC San Diego)
Privacy-preserving Analysis of Correlated data Kamalika Chaudhuri (UC San Diego) March 30, 2016 Abstract: Many modern machine learning applications involve private and sensitive data that are highly correlated. Examples are mining of time series of physical activity measurements, or minin
From playlist Nexus Trimester - 2016 - Secrecy and Privacy Theme
A short tutorial on differential privacy: Dr Borja Balle, Amazon Research
Differential privacy is a robust mathematical framework for designing privacy-preserving computations on sensitive data. In this tutorial we will cover the key definitions and intuitions behind differential privacy and introduce the core building blocks used by most differentially private
From playlist Turing Seminars
C68 The physics of damped motion
See how the graphs of damped motion changes with changes in mass, the spring constant, and the initial value constants. The equations tell us which parameters influence the period, frequency and amplitude of oscillation.
From playlist Differential Equations
Cecelia Clementi - Designing molecular models with machine learning and experimental data
Recorded 11 January 2023. Cecilia Clementi of Freie Universität Berlin presents "Designing molecular models with machine learning and experimental data" at IPAM's Explainable AI for the Sciences: Towards Novel Insights Workshop. Learn more online at: http://www.ipam.ucla.edu/programs/works
From playlist 2023 Explainable AI for the Sciences: Towards Novel Insights
Frank Noé: "Deep Generative Learning for Physics Many-Body Systems"
Machine Learning for Physics and the Physics of Learning 2019 Workshop I: From Passive to Active: Generative and Reinforcement Learning with Physics "Deep Generative Learning for Physics Many-Body Systems" Frank Noé, Freie Universität Berlin Institute for Pure and Applied Mathematics, UC
From playlist Machine Learning for Physics and the Physics of Learning 2019
Breaking the Communication-Privacy-Accuracy Trilemma
A Google TechTalk, 2020/7/29, presented by Ayfer Ozgur Aydin, Stanford University ABSTRACT: Two major challenges in distributed learning and estimation are 1) preserving the privacy of the local samples; and 2) communicating them efficiently to a central server, while achieving high accura
From playlist 2020 Google Workshop on Federated Learning and Analytics
Academic Keynote: Differentially Private Covariance-Adaptive Mean Estimation, Adam Smith (BU)
A Google TechTalk, presented by Adam Smith, 2021/11/9 ABSTRACT: Differentially Private Covariance-Adaptive Mean Estimation Covariance-adaptive mean estimation is a fundamental problem in statistics, where we are given n i.i.d. samples from a d-dimensional distribution with mean $\mu$ and
From playlist 2021 Google Workshop on Federated Learning and Analytics
How To Tackle Our Inverse Problem by Wolfgang Waltenberger
Discussion Meeting : Hunting SUSY @ HL-LHC (ONLINE) ORGANIZERS : Satyaki Bhattacharya (SINP, India), Rohini Godbole (IISc, India), Kajari Majumdar (TIFR, India), Prolay Mal (NISER-Bhubaneswar, India), Seema Sharma (IISER-Pune, India), Ritesh K. Singh (IISER-Kolkata, India) and Sanjay Kuma
From playlist HUNTING SUSY @ HL-LHC (ONLINE) 2021
Spring toggle mechanism enables to reach end positions of a lever quickly and holds it there firmly. The pink double crank represents action from outside.
From playlist Mechanisms
You may have seen some of my Q&A videos posted to YouTube and wondered where they come from. Some of them come from monthly Livestream Q&A sessions with people who support me on Patreon. (We call Patrons what they are: Community Builders.) Recently, Community Builders voted to open up thes
From playlist Bitcoin Q&A