The no-free-lunch theorems of supervised learning

Sterkenburg, Tom F.; Grünwald, Peter

doi:10.1007/s11229-021-03233-1

Cited by 38 publications

(21 citation statements)

References 58 publications

(76 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the authors suggest that future research should provide insights into the selection of different decomposition strategies. Note that this proposal is not intended to derive an optimal policy for all problems and scenarios, which is impossible according to the so-called “No Free Lunch Theorem” [ 113 ]. There is no “super algorithm” that works best for all problems.…”

Section: Discussionmentioning

confidence: 99%

Survey of Decomposition-Reconstruction-Based Hybrid Approaches for Short-Term Traffic State Forecasting

Chen

Wang

Hua

et al. 2022

Sensors

View full text Add to dashboard Cite

Traffic state prediction provides key information for intelligent transportation systems (ITSs) for proactive traffic management, the importance of which has become the reason for the tremendous number of research papers in this field. Over the last few decades, the decomposition-reconstruction (DR) hybrid models have been favored by numerous researchers to provide a more robust framework for short-term traffic state prediction for ITSs. This study surveyed DR-based works for short-term traffic state forecasting that were reported in the past circa twenty years, particularly focusing on how decomposition and reconstruction strategies could be utilized to enhance the predictability and interpretability of basic predictive models of traffic parameters. The reported DR-based models were classified and their applications in this area were scrutinized. Discussion and potential future directions are also provided to support more sophisticated applications. This work offers modelers suggestions and helps to choose appropriate decomposition and reconstruction strategies in their research and applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Survey of Decomposition-Reconstruction-Based Hybrid Approaches for Short-Term Traffic State Forecasting

Chen

Wang

Hua

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Thus, for every learning algorithm, there exists a task on which it fails, as no learning algorithm can generalize to all possible realities while having only observed some instances of the realities. To that end, NFL has also been discussed in relation to more fundamental problems in the philosophy of induction, e.g., in connection to Hume's problem of induction (Sterkenburg and Grünwald, 2021;Schurz, 2017) or Occham's razor (Lattimore and Hutter, 2013). Hume famously advanced skepticism against the very justification of induction, arguing that deductive reasoning alone cannot secure the validity of inductive inference; and neither can induction, due to circularity, provide non-deductive grounds for itself (Hume, 1739).…”

Section: Weak and Strong Sample Complexitymentioning

confidence: 99%

“…While they may obstruct the prospects of a perfect universal moral learner, it does not stop us from pursuing weaker yet reasonable alternatives that are practically viable. Instead of seeking a global and model-independent justification for why inductive inference seems to work, we can opt for local and model-relative justifications in order to explain why some learning algorithms work better than others (Sterkenburg and Grünwald, 2021). However, it should be stressed that any such alternative would inevitably entail some form of inductive bias; assumptions that we exploit to enable and foster learnability.…”

Section: Weak and Strong Sample Complexitymentioning

confidence: 99%

On the Computational Complexity of Ethics: Moral Tractability for Minds and Machines

Stenseke¹

2023

Preprint

View full text Add to dashboard Cite

Why should moral philosophers, moral psychologists, and machine ethicists care about computational complexity? Debates on whether artificial intelligence (AI) can or should be used to solve problems in ethical domains have mainly been driven by what AI can or cannot do in terms of human capacities. In this paper, we tackle the problem from the other end by exploring what kind of moral machines are possible based on what computational systems can or cannot do. To do so, we analyze normative ethics through the lens of computational complexity. First, we introduce computational complexity for the uninitiated reader and discuss how the complexity of ethical problems can be framed within Marr's three levels of analysis. We then study a range of ethical problems based on consequentialism, deontology, and virtue ethics, with the aim of elucidating the complexity associated with the problems themselves (e.g., due to combinatorics, uncertainty, strategic dynamics), the computational methods employed (e.g., probability, logic, learning), and the available resources (e.g., time, knowledge, learning). The results indicate that most problems the normative frameworks pose lead to tractability issues in every category analyzed. Our investigation also provides several insights about the computational nature of normative ethics, including the differences between rule-and outcome-based moral strategies, and the implementation-variance with regard to moral resources. We then discuss the consequences complexity results have for the prospect of moral machines in virtue of the trade-off between optimality and efficiency. Finally, we elucidate how computational complexity can be used to inform both philosophical and cognitive-psychological research on human morality by advancing the Moral Tractability Thesis (MTT).

show abstract

“…Wolpert has argued that NFL "can be viewed as a formalization and elaboration of concerns about the legitimacy of inductive inference, concerns that date back to David Hume (if not earlier)" (Wolpert, Wolpert (2012), 1). Philosophers have examined the relationship between NFL and inductive inference (Schurz, 2017) and have argued extensively against Wolpert's interpretation (Sterkenburg & Grünwald, 2021). Because Dotan's argument relies upon NFL, it would be useful to review the two main NFL results in search and optimization and in supervised learning in detail along with the philosophical discussion.…”

Section: No Free Lunch Theorem In Detailmentioning

confidence: 99%

No free theory choice from machine learning

Rushing

2022

Synthese

View full text Add to dashboard Cite

Ravit Dotan argues that a No Free Lunch theorem (NFL) from machine learning shows epistemic values are insufficient for deciding the truth of scientific hypotheses. She argues that NFL shows that the best case accuracy of scientific hypotheses is no more than chance. Since accuracy underpins every epistemic value, non-epistemic values are needed to assess the truth of scientific hypotheses. However, NFL cannot be coherently applied to the problem of theory choice. The NFL theorem Dotan’s argument relies upon is a member of a family of theorems in search, optimization, and machine learning. They all claim to show that if no assumptions are made about a search or optimization problem or learning situation, then the best case performance of an algorithm is that of random search or random guessing. A closer inspection shows that these theorems all rely upon assigning uniform probabilities over problems or learning situations, which is just the Principle of Indifference. A counterexample can be crafted that shows that NFL cannot be coherently applied across different descriptions of the same learning situation. To avoid this counterexample, Dotan needs to privilege some description of the learning situation faced by scientists. However, this means that NFL cannot be applied since an important assumption about the problem is being made. So Dotan faces a dilemma: either NFL leads to incoherent best-case partial beliefs or it is inapplicable to the problem of theory choice. This negative result has implications for the larger debate over theory choice.

show abstract

The no-free-lunch theorems of supervised learning

Cited by 38 publications

References 58 publications

Survey of Decomposition-Reconstruction-Based Hybrid Approaches for Short-Term Traffic State Forecasting

Survey of Decomposition-Reconstruction-Based Hybrid Approaches for Short-Term Traffic State Forecasting

On the Computational Complexity of Ethics: Moral Tractability for Minds and Machines

No free theory choice from machine learning

Contact Info

Product

Resources

About