1
|
Schmid M, Moravčík M, Burch N, Kadlec R, Davidson J, Waugh K, Bard N, Timbers F, Lanctot M, Holland GZ, Davoodi E, Christianson A, Bowling M. Student of Games: A unified learning algorithm for both perfect and imperfect information games. Sci Adv 2023; 9:eadg3256. [PMID: 37967182 PMCID: PMC10651118 DOI: 10.1126/sciadv.adg3256] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Received: 12/16/2022] [Accepted: 09/26/2023] [Indexed: 11/17/2023]
Abstract
Games have a long history as benchmarks for progress in artificial intelligence. Approaches using search and learning produced strong performance across many perfect information games, and approaches using game-theoretic reasoning and learning demonstrated strong performance for specific imperfect information poker variants. We introduce Student of Games, a general-purpose algorithm that unifies previous approaches, combining guided search, self-play learning, and game-theoretic reasoning. Student of Games achieves strong empirical performance in large perfect and imperfect information games-an important step toward truly general algorithms for arbitrary environments. We prove that Student of Games is sound, converging to perfect play as available computation and approximation capacity increases. Student of Games reaches strong performance in chess and Go, beats the strongest openly available agent in heads-up no-limit Texas hold'em poker, and defeats the state-of-the-art agent in Scotland Yard, an imperfect information game that illustrates the value of guided search, learning, and game-theoretic reasoning.
Collapse
Affiliation(s)
- Martin Schmid
- EquiLibre Technologies, Prague, Czechia
- Google Deepmind
| | | | - Neil Burch
- Google Deepmind
- Sony AI, New York, NY, USA
- Amii, Edmonton, Canada
| | - Rudolf Kadlec
- EquiLibre Technologies, Prague, Czechia
- Google Deepmind
| | | | | | - Nolan Bard
- Google Deepmind
- Sony AI, New York, NY, USA
| | | | - Marc Lanctot
- Google Deepmind
- Google Deepmind, Montreal, Canada
| | | | | | | | - Michael Bowling
- Google Deepmind
- Amii, Edmonton, Canada
- University of Alberta, Edmonton, Canada
| |
Collapse
|
2
|
Perolat J, De Vylder B, Hennes D, Tarassov E, Strub F, de Boer V, Muller P, Connor JT, Burch N, Anthony T, McAleer S, Elie R, Cen SH, Wang Z, Gruslys A, Malysheva A, Khan M, Ozair S, Timbers F, Pohlen T, Eccles T, Rowland M, Lanctot M, Lespiau JB, Piot B, Omidshafiei S, Lockhart E, Sifre L, Beauguerlange N, Munos R, Silver D, Singh S, Hassabis D, Tuyls K. Mastering the game of Stratego with model-free multiagent reinforcement learning. Science 2022; 378:990-996. [DOI: 10.1126/science.add4679] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/03/2022]
Abstract
We introduce DeepNash, an autonomous agent that plays the imperfect information game Stratego at a human expert level. Stratego is one of the few iconic board games that artificial intelligence (AI) has not yet mastered. It is a game characterized by a twin challenge: It requires long-term strategic thinking as in chess, but it also requires dealing with imperfect information as in poker. The technique underpinning DeepNash uses a game-theoretic, model-free deep reinforcement learning method, without search, that learns to master Stratego through self-play from scratch. DeepNash beat existing state-of-the-art AI methods in Stratego and achieved a year-to-date (2022) and all-time top-three ranking on the Gravon games platform, competing with human expert players.
Collapse
Affiliation(s)
| | | | | | | | | | | | | | | | | | | | | | | | | | - Zhe Wang
- DeepMind Technologies Ltd., London, UK
| | | | | | - Mina Khan
- DeepMind Technologies Ltd., London, UK
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
Collapse
|
3
|
|
4
|
Bard N, Foerster JN, Chandar S, Burch N, Lanctot M, Song HF, Parisotto E, Dumoulin V, Moitra S, Hughes E, Dunning I, Mourad S, Larochelle H, Bellemare MG, Bowling M. The Hanabi challenge: A new frontier for AI research. ARTIF INTELL 2020. [DOI: 10.1016/j.artint.2019.103216] [Citation(s) in RCA: 33] [Impact Index Per Article: 8.3] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/29/2022]
|
5
|
Abstract
The CFR+ algorithm for solving imperfect information games is a variant of the popular CFR algorithm, with faster empirical performance on a range of problems. It was introduced with a theoretical upper bound on solution error, but subsequent work showed an error in one step of the proof. We provide updated proofs to recover the original bound.
Collapse
|
6
|
Colby J, Bradley J, Durran-Dyer D, Read G, Griffiths C, Burch N. A model of an effective supportive Community Nursing Service for Tube fed patients. Clin Nutr ESPEN 2019. [DOI: 10.1016/j.clnesp.2018.12.029] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/25/2022]
|
7
|
Hollingworth T, Oke S, Akbar T, McKee R, Rochford A, Relph WL, Burch N, Smith T, Gabe S. The composition of nutrition support teams in the UK. Clin Nutr ESPEN 2019. [DOI: 10.1016/j.clnesp.2018.12.051] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/26/2022]
|
8
|
Ford R, Burch N. A pilot of the malnutrition universal screening tool (‘MUST’) in a general outpatient department: A service user questionnaire. Clin Nutr ESPEN 2018. [DOI: 10.1016/j.clnesp.2018.09.015] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/28/2022]
|
9
|
Kilbane T, Whitefield R, Colby J, Burch N. Monitoring compliance with nasogastric (NG) tube management: A single centre experience. Clin Nutr ESPEN 2018. [DOI: 10.1016/j.clnesp.2018.09.053] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/28/2022]
|
10
|
Bridgestock H, Bell C, Whitefield R, Sothi S, Burch N. Home Parenteral Nutrition (HPN) in palliative oncology patients: A UK single centre case series. Clin Nutr ESPEN 2017. [DOI: 10.1016/j.clnesp.2017.07.033] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
|
11
|
Mcfarlane M, Hammond C, Roper T, Mukarati J, Gordon V, Burch N. PTU-111 Nutrition in cirrhosis: why we must save them from turning to (l)dust? Nutrition 2017. [DOI: 10.1136/gutjnl-2017-314472.206] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/04/2022]
|
12
|
Moravčík M, Schmid M, Burch N, Lisý V, Morrill D, Bard N, Davis T, Waugh K, Johanson M, Bowling M. DeepStack: Expert-level artificial intelligence in heads-up no-limit poker. Science 2017; 356:508-513. [PMID: 28254783 DOI: 10.1126/science.aam6960] [Citation(s) in RCA: 295] [Impact Index Per Article: 42.1] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/03/2017] [Accepted: 02/21/2017] [Indexed: 11/02/2022]
Abstract
Artificial intelligence has seen several breakthroughs in recent years, with games often serving as milestones. A common feature of these games is that players have perfect information. Poker, the quintessential game of imperfect information, is a long-standing challenge problem in artificial intelligence. We introduce DeepStack, an algorithm for imperfect-information settings. It combines recursive reasoning to handle information asymmetry, decomposition to focus computation on the relevant decision, and a form of intuition that is automatically learned from self-play using deep learning. In a study involving 44,000 hands of poker, DeepStack defeated, with statistical significance, professional poker players in heads-up no-limit Texas hold'em. The approach is theoretically sound and is shown to produce strategies that are more difficult to exploit than prior approaches.
Collapse
Affiliation(s)
- Matej Moravčík
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada.,Department of Applied Mathematics, Charles University, Prague, Czech Republic
| | - Martin Schmid
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada.,Department of Applied Mathematics, Charles University, Prague, Czech Republic
| | - Neil Burch
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Viliam Lisý
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada.,Department of Computer Science, Faculty of Electrical Engineering, Czech Technical University, Prague, Czech Republic
| | - Dustin Morrill
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Nolan Bard
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Trevor Davis
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Kevin Waugh
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Michael Johanson
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada
| | - Michael Bowling
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada.
| |
Collapse
|
13
|
|
14
|
Affiliation(s)
- Robert C. Holte
- Computing Science Department, University of Alberta, Edmonton, AB, Canada. E-mails: {holte, nburch}@cs.ualberta.ca
| | - Neil Burch
- Computing Science Department, University of Alberta, Edmonton, AB, Canada. E-mails: {holte, nburch}@cs.ualberta.ca
| |
Collapse
|
15
|
Abstract
Korf, Reid, and Edelkamp introduced a formula to predict the number of nodes IDA* will expand on a single iteration for a given consistent heuristic, and experimentally demonstrated that it could make very accurate predictions. In this paper we show that, in addition to requiring the heuristic to be consistent, their formula's predictions are accurate only at levels of the brute-force search tree where the heuristic values obey the unconditional distribution that they defined and then used in their formula. We then propose a new formula that works well without these requirements, i.e., it can make accurate predictions of IDA*'s performance for inconsistent heuristics and if the heuristic values in any
level do not obey the unconditional distribution. In order to achieve this we introduce the conditional distribution of heuristic values which is a generalization of their unconditional heuristic distribution. We also provide extensions of our formula that handle individual start states and the augmentation of IDA* with bidirectional pathmax (BPMX), a technique for propagating heuristic values when inconsistent heuristics are used. Experimental results demonstrate the accuracy of our new method and all its variations.
Collapse
|
16
|
Biran M, Burch N. Emtional Intelligence, Depression, and Antidepressants. Eur Psychiatry 2009. [DOI: 10.1016/s0924-9338(09)70849-1] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/25/2022] Open
Abstract
Aims:Emotional intelligence is the measure of the comprehension of one's emotional state and the capability to manage emotions, to recognize emotions in others and to react to those emotions appropriately (Salovey & Mayer, 1990).In previous studies a strong correlation was found between low emotional intelligence and depression. (e.g.,Berrocal, Alcaide, Extremera, & Pizarro, 2006).Since antidepressants attempt to solve depression by numbing patients’ emotions and making the emotions more controllable (Arce et al, 2008; Davison, 2005), it was our hypothesis that antidepressants will lower emotional intelligence.Method:A package of questionnaires was distributed to 195 undergraduate students. Included in the package was a questionnaire asking subjects whether depressed, and were they taking antidepressants. Also included were the SCL-90 used to assess emotional distress (Derogatis,1985), the Beck Depression Inventory used to assess level of depression (Beck, 1981), and the Schutte Emotional Intelligence Scale used to evaluate level of emotional intelligence.Results:The results not only agreed with the findings of previous research linking low levels of emotional intelligence to depression, but also supported the hypothesis that the use of antidepressants would be linked to lower levels of emotional intelligence (r (191)= -.158, p< .029) This is a powerful finding, considering that the sample size was rather small. Only about 15% of the sample reported being prescribed antidepressants to treat their depression.Conclusions:The results of this study suggest that taking antidepressants as a form of treatment for depression might impact negatively on patients’ ability to handle their emotions.
Collapse
|
17
|
Abstract
BACKGROUND Patients with Barrett's oesophagus are at increased risk of oesophageal adenocarcinoma. Observational studies have suggested increase in overall mortality also but data are conflicting. AIM To assess the cause of death in patients with Barrett's oesophagus compared with the general population. METHODS Patients with Barrett's oesophagus were identified retrospectively in four hospitals in Leicestershire, UK using electronic endoscopy and histopathology records from 1997 to 2003. Data on deaths from this cohort of patients were identified through the Office of National Statistics and compared with age- and gender-adjusted mortality in the Leicestershire region. RESULT In all, 1272 Barrett's patients were identified with 245 deaths in this cohort. Overall mortality was found to be increased [male standardized mortality ratio (SMR) = 552, 95% CI = 466-638; female SMR 455, 95% CI = 357-552]. The main disease areas that were responsible for this increase were oesophageal adenocarcinoma (n = 25, male SMR = 2171, 95% CI = 991-3351; female SMR = 1300, 95% CI = 26-2574), bronchopneumonia (n = 70, male SMR = 146, 95% CI = 55-236; female SMR = 436, 95% CI = 272-601) and ischaemic heart disease (n = 51, male SMR = 186, 95% CI = 97-2748; female SMR = 205, 95% CI = 105-306). CONCLUSIONS Patients with Barrett's oesophagus die more commonly of bronchopneumonia and ischaemic heart disease compared with oesophageal adenocarcinoma, and overall mortality in this group may be increased.
Collapse
Affiliation(s)
- P Moayyedi
- Division of Gastroenterology, Department of Medicine, McMaster University, Hamilton, ON, Canada.
| | | | | | | | | | | | | |
Collapse
|
18
|
Abstract
The game of checkers has roughly 500 billion billion possible positions (5 x 10(20)). The task of solving the game, determining the final result in a game with no mistakes made by either player, is daunting. Since 1989, almost continuously, dozens of computers have been working on solving checkers, applying state-of-the-art artificial intelligence techniques to the proving process. This paper announces that checkers is now solved: Perfect play by both sides leads to a draw. This is the most challenging popular game to be solved to date, roughly one million times as complex as Connect Four. Artificial intelligence technology has been used to generate strong heuristic-based game-playing programs, such as Deep Blue for chess. Solving a game takes this to the next level by replacing the heuristics with perfection.
Collapse
Affiliation(s)
- Jonathan Schaeffer
- Department of Computing Science, University of Alberta, Edmonton, Alberta T6G 2E8, Canada.
| | | | | | | | | | | | | | | |
Collapse
|
19
|
Billings D, Davidson A, Schauenberg T, Burch N, Bowling M, Holte R, Schaeffer J, Szafron D. Game-Tree Search with Adaptation in Stochastic Imperfect-Information Games. Computers and Games 2006. [DOI: 10.1007/11674399_2] [Citation(s) in RCA: 16] [Impact Index Per Article: 0.9] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 01/03/2023]
|
20
|
Agaba AE, Hardiment K, Burch N, Imray CHE. An Audit of Vascular Surgical Intervention for Complications of Cardiovascular Angiography in 2324 Patients from a Single Center. Ann Vasc Surg 2004; 18:470-3. [PMID: 15164261 DOI: 10.1007/s10016-004-0041-6] [Citation(s) in RCA: 5] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/21/2022]
Abstract
This study examines the incidence and outcome of complications requiring surgical intervention in a major vascular unit serving interventional radiology and interventional cardiology. Between April 2000 and March 2001, 2324 patients underwent angiographic examinations (male/female = 1579:745, mean age = 68 years, range 45-88). In non-stent patients, a 4-or 5-mm French (4-mm F, 5-mm F) guage nonheparinized arterial catheter was used, and in patients requiring stents a 6- or 7-mm French guage catheter was used. Pressure was applied to the puncture site for up to 6 min. Fifteen complications requiring vascular surgical procedure were recorded during in-hospital follow-up (9 males, 6 females). Our early operative (30-day) mortality rate was 0.086%. Although the number of major complications requiring surgical intervention after interventional or diagnostic cardiovascular radiology is diminishing, vigilance in these cases is still required. Where possible, a small catheter with a J-shaped guidewire should be used and prolonged compression should be brought to bear on the puncture site.
Collapse
Affiliation(s)
- A E Agaba
- Department of Vascular Surgery, Walsgrave Hospital NHS Trust, Coventry, UK.
| | | | | | | |
Collapse
|
21
|
Burch N. VA-community nursing home care program called outstanding success. Mod Nurs Home 1972; 28:38-9. [PMID: 4623656] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [MESH Headings] [Subscribe] [Scholar Register] [Indexed: 01/11/2023]
|