ML Meetup: The Animal-AI Olympics & Amortized Monte Carlo Integration

Man Technology Team

Man Technology

At Man Group, we believe in the Python Ecosystem and have been trading Machine Learning based systems since early 2014.

July 2019

To give back and strengthen London’s Python and Machine Learning Communities, we sponsor and support the PyData and Machine Learning London Meetups.

In July, we had the pleasure of welcoming Matt Crosby, postdoctoral researcher at the Leverhulme Centre for the Future of Intelligence at Imperial College London, and Tom Rainforth, postdoctoral researcher in statistical ML at the University of Oxford, to the London Machine Learning Meetup.

 

The Animal-AI Olympics: Translating animal cognition tasks for AI - Matthew Crosby

The Animal-AI Olympics is an AI competition that translates experiments from animal cognition and presents them as a challenge for AI. Animal cognition tasks are designed to test for specific capabilities, so that solving them demonstrates that an animal understands certain properties of its environment or has certain cognitive abilities. By presenting these tasks to AI, the aim is to find out how to build intelligent agents with similar understanding or cognitive abilities. The translation from animals to AI is not trivial. Animals come to the experiments as curious food-motivated explorers, and part of the challenge is to encourage AI systems to be built in a similar way. A key element of many animal cognition tasks requires testing animal responses to first presentation of a novel situation, therefore ruling out the possibility of solutions involving repetition of previously learned behaviour. This goes against standard practice in ML, where training and test sets are often drawn from the same probability distribution.

 

Matthew Crosby

Matt Crosby is currently a postdoc researcher at the Leverhulme Centre for the Future of Intelligence at Imperial College London. He has a PhD in AI, and more recently has been working on artificial cognition. He is currently running the Animal-AI Olympics, an AI competition based on animal cognition tasks. More generally he is interested in understanding the links between (artificial) intelligence and cognition & exploring the space of possible minds.

 

Amortized Monte Carlo Integration - Tom Rainforth

Current approaches to amortizing Bayesian inference focus solely on approximating the posterior distribution. Typically, this approximation is, in turn, used to calculate expectations for one or more target functions - a computational pipeline which is inefficient when the target function(s) are known upfront. In this work, we address this inefficiency by introducing AMCI, a method for amortizing Monte Carlo integration directly. AMCI operates similarly to amortized inference but produces three distinct amortized proposals, each tailored to a different component of the overall expectation calculation. At runtime, samples are produced separately from each amortized proposal, before being combined to an overall estimate of the expectation. We show that while existing approaches are fundamentally limited in the level of accuracy they can achieve, AMCI can theoretically produce arbitrarily small errors for any integrable target function using only a single sample from each proposal at runtime. We further show that it is able to empirically outperform the theoretically optimal self-normalized importance sampler on a number of example problems. AMCI allows not only for amortizing over datasets but also amortizing over target functions.

 

Tom Rainforth

Tom Rainforth is a postdoc researcher in statistical ML at the University of Oxford. His research covers a wide range of topics including probabilistic programming, Monte Carlo methods, variational inference, Bayesian optimization, experimental design, and deep generative models. In addition to his current role as a postdoc working with Yee Whye Teh, he co-supervises a small group of his own students. Prior to this, Tom undertook a D.Phil under the supervision of Frank Wood, culminating in the thesis “Automating inference, learning & design using probabilistic programming.”

 

I am interested in other Tech Articles.

To receive e-mail alerts whenever new Tech Articles or Events are posted on this site, please subscribe below.

Subscribe

 

Find out more about Technology at Man Group

重要資料

In the case of hypothetical results:

Hypothetical Results are calculated in hindsight, invariably show positive rates of return, and are subject to various modeling assumptions, statistical variances and interpretational differences. No representation is made as to the reasonableness or accuracy of the calculations or assumptions made or that all assumptions used in achieving the results have been utilized equally or appropriately, or that other assumptions should not have been used or would have been more accurate or representative. Changes in the assumptions would have a material impact on the Hypothetical Results and other statistical information based on the Hypothetical Results.

The Hypothetical Results have other inherent limitations, some of which are described below. They do not involve financial risk or reflect actual trading by an Investment Product, and therefore do not reflect the impact that economic and market factors, including concentration, lack of liquidity or market disruptions, regulatory (including tax) and other conditions then in existence may have on investment decisions for an Investment Product. In addition, the ability to withstand losses or to adhere to a particular trading program in spite of trading losses are material points which can also adversely affect actual trading results. Since trades have not actually been executed, Hypothetical Results may have under or over compensated for the impact, if any, of certain market factors. There are frequently sharp differences between the Hypothetical Results and the actual results of an Investment Product. No assurance can be given that market, economic or other factors may not cause the Investment Manager to make modifications to the strategies over time. There also may be a material difference between the amount of an Investment Product’s assets at any time and the amount of the assets assumed in the Hypothetical Results, which difference may have an impact on the management of an Investment Product. Hypothetical Results should not be relied on, and the results presented in no way reflect skill of the investment manager. A decision to invest in an Investment Product should not be based on the Hypothetical Results.

No representation is made that an Investment Product’s performance would have been the same as the Hypothetical Results had an Investment Product been in existence during such time or that such investment strategy will be maintained substantially the same in the future; the Investment Manager may choose to implement changes to the strategies, make different investments or have an Investment Product invest in other investments not reflected in the Hypothetical Results or vice versa. To the extent there are any material differences between the Investment Manager’s management of an Investment Product and the investment strategy as reflected in the Hypothetical Results, the Hypothetical Results will no longer be as representative and their illustration value will decrease substantially. No representation is made that an Investment Product will or is likely to achieve its objectives or results comparable to those shown, including the Hypothetical Results, or will make any profit or will be able to avoid incurring substantial losses. Past performance is not indicative of future results and simulated results in no way reflect upon the manger’s skill or ability.

 

Important Information

This information is communicated and/or distributed by the relevant Man entity identified below (collectively the "Company") subject to the following conditions and restriction in their respective jurisdictions.

Opinions expressed are those of the author and may not be shared by all personnel of Man Group plc (‘Man’). These opinions are subject to change without notice, are for information purposes only and do not constitute an offer or invitation to make an investment in any financial instrument or in any product to which the Company and/or its affiliates provides investment advisory or any other financial services. Any organisations, financial instrument or products described in this material are mentioned for reference purposes only which should not be considered a recommendation for their purchase or sale. Neither the Company nor the authors shall be liable to any person for any action taken on the basis of the information provided. Some statements contained in this material concerning goals, strategies, outlook or other non-historical matters may be forward-looking statements and are based on current indicators and expectations. These forward-looking statements speak only as of the date on which they are made, and the Company undertakes no obligation to update or revise any forward-looking statements. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from those contained in the statements. The Company and/or its affiliates may or may not have a position in any financial instrument mentioned and may or may not be actively trading in any such securities. Unless stated otherwise all information is provided by the Company. Past performance is not indicative of future results.

Unless stated otherwise this information is communicated by the relevant entity listed below.

Australia: To the extent this material is distributed in Australia it is communicated by Man Investments Australia Limited ABN 47 002 747 480 AFSL 240581, which is regulated by the Australian Securities & Investments Commission ('ASIC'). This information has been prepared without taking into account anyone’s objectives, financial situation or needs.

Austria/Germany/Liechtenstein: To the extent this material is distributed in Austria, Germany and/or Liechtenstein it is communicated by Man (Europe) AG, which is authorised and regulated by the Liechtenstein Financial Market Authority (FMA). Man (Europe) AG is registered in the Principality of Liechtenstein no. FL-0002.420.371-2. Man (Europe) AG is an associated participant in the investor compensation scheme, which is operated by the Deposit Guarantee and Investor Compensation Foundation PCC (FL-0002.039.614-1) and corresponds with EU law. Further information is available on the Foundation's website under www.eas-liechtenstein.li.

European Economic Area: Unless indicated otherwise this material is communicated in the European Economic Area by Man Asset Management (Ireland) Limited (‘MAMIL’) which is registered in Ireland under company number 250493 and has its registered office at 70 Sir John Rogerson's Quay, Grand Canal Dock, Dublin 2, Ireland. MAMIL is authorised and regulated by the Central Bank of Ireland under number C22513.

Hong Kong SAR: To the extent this material is distributed in Hong Kong SAR, this material is communicated by Man Investments (Hong Kong) Limited and has not been reviewed by the Securities and Futures Commission in Hong Kong.

Japan: To the extent this material is distributed in Japan it is communicated by Man Group Japan Limited, Financial Instruments Business Operator, Director of Kanto Local Finance Bureau (Financial instruments firms) No. 624 for the purpose of providing information on investment strategies, investment services, etc. provided by Man Group, and is not a disclosure document based on laws and regulations. This material can only be communicated only to professional investors (i.e. specific investors or institutional investors as defined under Financial Instruments Exchange Law) who may have sufficient knowledge and experience of related risks.

Switzerland: To the extent this material is made available in Switzerland the communicating entity is:

  • For Clients (as such term is defined in the Swiss Financial Services Act): Man Investments (CH) AG, Huobstrasse 3, 8808 Pfäffikon SZ, Switzerland. Man Investment (CH) AG is regulated by the Swiss Financial Market Supervisory Authority (‘FINMA’); and
  • For Financial Service Providers (as defined in Art. 3 d. of FINSA, which are not Clients): Man Investments AG, Huobstrasse 3, 8808 Pfäffikon SZ, Switzerland, which is regulated by FINMA.

United Kingdom: Unless indicated otherwise this material is communicated in the United Kingdom by Man Solutions Limited ('MSL') which is a private limited company registered in England and Wales under number 3385362. MSL is authorised and regulated by the UK Financial Conduct Authority (the 'FCA') under number 185637 and has its registered office at Riverbank House, 2 Swan Lane, London, EC4R 3AD, United Kingdom.

United States: To the extent this material is distributed in the United States, it is communicated and distributed by Man Investments, Inc. (‘Man Investments’). Man Investments is registered as a broker-dealer with the SEC and is a member of the Financial Industry Regulatory Authority (‘FINRA’). Man Investments is also a member of the Securities Investor Protection Corporation (‘SIPC’). Man Investments is a wholly owned subsidiary of Man Group plc. The registration and memberships described above in no way imply a certain level of skill or expertise or that the SEC, FINRA or the SIPC have endorsed Man Investments. Man Investments Inc, 1345 Avenue of the Americas, 21st Floor, New York, NY 10105.

This material is proprietary information and may not be reproduced or otherwise disseminated in whole or in part without prior written consent. Any data services and information available from public sources used in the creation of this material are believed to be reliable. However accuracy is not warranted or guaranteed. © Man 2024