non-stationarity
A property of a system whose underlying statistical structure changes over time, so that models built on past data no longer describe the current or future world.
As defined in If Life Is Random, Is It Meaningless? (with Brian Klaas)
Other definitions
Property of data whose statistical relationships change over time, which undermines learning from history.
The Past, The Present & Future of QuantWhen the statistical properties of data change over time, so patterns learned from history may not hold in the future.
Karen Karniol-Tambour - Macro Headwinds vs. Tech Tailwinds - [Invest Like the Best, EP.329]
Where it comes up · 6
- Making Decisions with Data3 claims
- Upside Down Markets2 claims
- If Life Is Random, Is It Meaningless? (with Brian Klaas)2 claims
- Elon Musk And Neuralink Team2 claims
- Ex-Google CEO: What Artificial Superintelligence Will Actually Look Like w/ Eric Schmidt & Dave B2 claims
- Tim Cook’s Dashboard1 claim
Most important claims · 10 of 12Search all claims for “non-stationarity” →
- Top sports bettors have little durable edge because they must keep changing their approach, and static historical models for the NBA fail as rest management, rules, and schedules shift.Making Decisions with Data
- There is little historical data on fiscal stimulus and inflation, and the relationship is non-stationary with many variables, so simple empirical arguments that a given amount of stimulus produces a given amount of inflation are hard to make.Upside Down Markets
- By becoming so consumed with modeling the world we have forgotten how reality actually works; models necessarily omit the contingent events (like officials' vacation histories) that are sometimes the true main cause of outcomes, so the tidy models reflect back a false image of a controllable world.If Life Is Random, Is It Meaningless? (with Brian Klaas)
- The growth of a business and its segments is relatively stationary, so machine learning works well on it, while asset prices are non-stationary, driven by mood and regime, making price prediction with machine learning a poor application.Tim Cook’s Dashboard
- The baseline firing rate of neurons shifts day to day (like weighing rice in a different pot each day), and since the brain represents motor intention via rate coding—the delta between baseline and modulated firing—this baseline drift is the first-order cause of cursor bias and the downstream non-stationarity problem that requires recalibration.Elon Musk And Neuralink Team
- Modeling the future from past data gets you into trouble when the world has actually changed—the problem of non-stationarity—because you are effectively modeling a different world from the one in which your data were collected.If Life Is Random, Is It Meaningless? (with Brian Klaas)
- Great scientists historically were experts in one field who discovered patterns from that field that could apply to completely unrelated fields, creating breakthroughs—a capability called non-stationarity that current AI models cannot perform.Ex-Google CEO: What Artificial Superintelligence Will Actually Look Like w/ Eric Schmidt & Dave B
- Scaling electrode channel count yields a logarithmic improvement in decode quality, increases the number of distinct actions a user can perform (since each channel maps to imagined movements), and crucially should improve system reliability against neural non-stationarity without any decoder changes, because as channel count grows the relative importance of any single channel's drift or noise diminishes.Elon Musk And Neuralink Team
- A conservative velocity-based analysis suggested that very large amounts of stimulus would produce less inflation than expected, making Livermore somewhat less worried about inflation than he was before, though still concerned.Upside Down Markets
- The 'non-stationarity problem' is the algorithmic challenge of teaching AI systems to apply old rules to new domains when 'rules keep changing'—current reward functions are straightforward ('beat the human, beat the question') but non-stationary learning requires fundamentally different architecture, and research is underway but outcome uncertain.Ex-Google CEO: What Artificial Superintelligence Will Actually Look Like w/ Eric Schmidt & Dave B