SlideShare a Scribd company logo
Proving theorems: still a major test for AI
Dr Sean Holden
University of Cambridge
Department of Computer Science and Technology
and
Trinity College
Codiax 2019
How did I get here?
Machine learning research since 1989:
• Initially on computational learning theory, particularly sample complexity of
the cross-validation and other estimates.
• Applications in drug design with Glaxo-SmithKline and others.
• Organelle proteomics in collaboration with the Cambridge Department of Bio-
chemistry.
• Various other areas in Bayesian inference, survey propagation, protein interac-
tions, gene expression and so on.
Most recently: I’ve been working on the question of…
How did I get here?
How can machine learning be applied to proving mathematical theorems?
How did I get here? Principia Mathematica…
Logicism: is all mathematics just logic?
• Alfred North Whitehead and Bertrand Russell attempted to demonstrate this.
• Three volumes, published in 1910, 1912 and 1913.
• Initially, very basic logical axioms. Then a systematic development.
• Proofs in their system can easily be searched for by an automated theorem
prover.
Famously, the important theorem
1 + 1 = 2
appears on page 83 of Volume 2.
Take-home message: being precise about mathematics is hard!
However, this has formed a fundamental challenge in AI since its earliest days:
the General Problem Solver, Herbert A. Simon, Cliff Shaw and Allen Newell, 1959.
Proofs in propositional/Boolean logic
Propositional logic:
• Variables can be true (T) or false (F).
• I have connectives ∧ (AND), ∨ (OR) and ¬ (NOT).
1. ¬x is T when x is F, and F when x is T.
2. x1 ∧ x2 is T precisely when both x1 and x2 are T.
3. x1 ∨ x2 is T precisely when x1 or x2 (or both) are T.
• For example, the formula (x1 ∨ x2) ∧ x3 is T when x1 = T, x2 = F and x3 = T.
Here is the game: I give you a HUGE formula with n variables:
(x1 ∨ x2 ∨ ¬x3) ∧ (¬x2 ∨ · · · ∨ x10) ∧ · · · ∧ (¬x1,000,000)
and you have to tell me whether or not there is a way of setting the variables to
make the formula T.
This is the satisfiability (SAT) problem.
There is no known method that can guarantee to be better than testing all 2n
possibilities!
Proofs in first-order logic
Am I a WEIRDO?
∀x . Pierced(x) ∧ Male(x) −→ Weirdo(x)
Pierced(sean)
Male(sean)
Does Weirdo(sean) follow?
We should probably agree that it does. (Hooray!)
First-order logic
The reason this is hard is simple:
{W(sean)}
{}
{P(sean)} {M(sean)}
x = sean
The set of clauses grows
There is a choice of which pair of clauses to resolve
{¬W(sean)}{¬P(x), ¬M(x), W(x)}
{¬M(sean), W(sean)}
And it’s worse than it seems:
• The full problem is semidecidable.
• (You can prove a valid theorem, but you might not be able to disprove an invalid
one.)
• The propositional subproblem is NP-complete.
Have computers managed to prove anything? The Robbins conjecture
So automated theorem proving is known to be hard. But has it been used to prove
any original mathematics?
Robbins algebras:
• Commutative:
∀a, b . a ∨ b = b ∨ a
• Associative:
∀a, b, c . (a ∨ b) ∨ c = a ∨ (b ∨ c)
• Robbins axiom
∀a, b . ¬(¬(a ∨ b) ∨ ¬(a ∨ ¬b)) = a
Robbins conjecture:
All Robbins algebras are Boolean.
The proof was found by William McClune using the EQP theorem prover. (Jour-
nal of Automated Reasoning, 1997.)
So what’s the problem?
So, what’s the problem?
Although the Robbins conjecture represents a proof of a genuine theorem by an
automated theorem prover:
• It is something of an isolated result.
• It essentially represents an application of brute-force search.
• The proof has little if any resemblance to the kind of proof a person might
construct.
• It’s not enough to impress the mathematicians.
What will it take to impress them?
We could aim to prove something BIG!
We could aim to prove multiple results with the same system without human
guidance. (Domain expertise used to tune the prover.)
First-order logic + machine learning
There’s a growing interest in improving ATPs by adding machine learning to
them.
I’ve been interested in this for some time:
• The E theorem prover is a state-of-the-art prover for predicate logic.
• One way it tries to address the problem is to use features of the theorem of
interest to choose from 82 heuristics for choosing what to resolve.
• Can we improve on this?
We found that, yes, we could. (Bridge, Holden and Paulson, Journal of Automated
Reasoning, 2014).
Specifically, applying support vector machines and Gaussian process classifiers we
achieved comparable performace with only 5 heuristics.
Take-home message: machine learning can provide useful guidance for automated
theorem provers.
This proof is quite big…
The classification theorem for finite simple groups
Every finite, simple group is isomorphic to one of:
• a cyclic group with prime order
• an alternating group with degree ≥ 5
• a group of Lie type
• one of 26 sporadic groups.
This proof is quite big…
The proof of the theorem was announced in the early 1980s. BUT it was, er, quite
big:
• Estimated at > 10, 000 pages long.
• After announcement, it turned out there was a gap related to quasithin groups…
• … which took another 7 years and 1221 pages.
• There is an effort at present to shorten the proof .
• As of 2019 this is 8 volumes, with another 5 likely to be needed.
QUESTION: Is this proof correct?
(The mathematicians don’t seem too worried, but I’m not so sure…)
Proof checking: formalising mathematical results
Systems such as Isabelle provide a means of constructing formally verified proofs:
https://www.cl.cam.ac.uk/research/hvg/Isabelle/
Such systems are often presented as proof assistants.
There is a body of formalized mathematics: the Archive of Formal Proofs (AFP):
https://www.isa-afp.org/
This provides us with another target for machine learning:
Lemma recommendation
Machine learning for selecting premises
Current work on lemma recommendation (Słowik, Mangla, Jamnik, Holden, Paul-
son, AAAI 2019).
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x
-2
-1
0
1
2
3
4
5
f(x)
Squared exponential prior, l = 1/2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x
-3
-2
-1
0
1
2
3
4
5
f(x)
γ-exponential prior, l = γ = 1/2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x
-2
-1
0
1
2
3
4
5
6f(x) Rational quadratic prior, l = a = 3
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x
-2
-1
0
1
2
3
4
5
f(x)
Neural network prior, σ1 = 15, σ2 = 10
• We build on the existing Sumo Inference Engine (SInE).
• We use Bayesian optimization to tune SInE.
• The AFP is a natural source of training data.
Mathematical language is tricky
Question: why don’t more mathematicians use proof assistants?
Answer: mathematicians don’t necessarily want to learn a new formal language.
CHALLENGE
Is there a way of writing mathematics that looks natural while being formally
verifiable?
It turns out that this is REALLY HARD!
See for example The Language of Mathematics: A Linguistic and Philosophical
Investigation, by Mohan Ganesalingam, Springer, 2013.

More Related Content

What's hot

Mixed Effects Models - Random Slopes
Mixed Effects Models - Random SlopesMixed Effects Models - Random Slopes
Mixed Effects Models - Random Slopes
Scott Fraundorf
 
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
Navodaya Institute of Technology
 
Mixed Effects Models - Random Intercepts
Mixed Effects Models - Random InterceptsMixed Effects Models - Random Intercepts
Mixed Effects Models - Random Intercepts
Scott Fraundorf
 
Continual Learning with Deep Architectures - Tutorial ICML 2021
Continual Learning with Deep Architectures - Tutorial ICML 2021Continual Learning with Deep Architectures - Tutorial ICML 2021
Continual Learning with Deep Architectures - Tutorial ICML 2021
Vincenzo Lomonaco
 
Big-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunitiesBig-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunities
台灣資料科學年會
 
An alternative learning experience in transition level mathematics
An alternative learning experience in transition level mathematicsAn alternative learning experience in transition level mathematics
An alternative learning experience in transition level mathematics
Dann Mallet
 
Instrumentation in Teaching Mathematics
Instrumentation in Teaching MathematicsInstrumentation in Teaching Mathematics
Instrumentation in Teaching Mathematics
MaryAlyssaCacha
 
Data simulation basics
Data simulation basicsData simulation basics
Data simulation basics
Dorothy Bishop
 
NAACL HLT 2010 d-Confidence
NAACL HLT 2010 d-ConfidenceNAACL HLT 2010 d-Confidence
NAACL HLT 2010 d-ConfidenceNunoEscudeiro
 
Mixed Effects Models - Fixed Effects
Mixed Effects Models - Fixed EffectsMixed Effects Models - Fixed Effects
Mixed Effects Models - Fixed Effects
Scott Fraundorf
 
Presentatie researchED Amsterdam
Presentatie researchED AmsterdamPresentatie researchED Amsterdam
Presentatie researchED Amsterdam
Christian Bokhove
 
Simulating data to gain insights into power and p-hacking
Simulating data to gain insights intopower and p-hackingSimulating data to gain insights intopower and p-hacking
Simulating data to gain insights into power and p-hacking
Dorothy Bishop
 
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
taxonbytes
 
Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401butest
 

What's hot (14)

Mixed Effects Models - Random Slopes
Mixed Effects Models - Random SlopesMixed Effects Models - Random Slopes
Mixed Effects Models - Random Slopes
 
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
An algorithm for solving unbalanced intuitionistic fuzzy assignment problem u...
 
Mixed Effects Models - Random Intercepts
Mixed Effects Models - Random InterceptsMixed Effects Models - Random Intercepts
Mixed Effects Models - Random Intercepts
 
Continual Learning with Deep Architectures - Tutorial ICML 2021
Continual Learning with Deep Architectures - Tutorial ICML 2021Continual Learning with Deep Architectures - Tutorial ICML 2021
Continual Learning with Deep Architectures - Tutorial ICML 2021
 
Big-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunitiesBig-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunities
 
An alternative learning experience in transition level mathematics
An alternative learning experience in transition level mathematicsAn alternative learning experience in transition level mathematics
An alternative learning experience in transition level mathematics
 
Instrumentation in Teaching Mathematics
Instrumentation in Teaching MathematicsInstrumentation in Teaching Mathematics
Instrumentation in Teaching Mathematics
 
Data simulation basics
Data simulation basicsData simulation basics
Data simulation basics
 
NAACL HLT 2010 d-Confidence
NAACL HLT 2010 d-ConfidenceNAACL HLT 2010 d-Confidence
NAACL HLT 2010 d-Confidence
 
Mixed Effects Models - Fixed Effects
Mixed Effects Models - Fixed EffectsMixed Effects Models - Fixed Effects
Mixed Effects Models - Fixed Effects
 
Presentatie researchED Amsterdam
Presentatie researchED AmsterdamPresentatie researchED Amsterdam
Presentatie researchED Amsterdam
 
Simulating data to gain insights into power and p-hacking
Simulating data to gain insights intopower and p-hackingSimulating data to gain insights intopower and p-hacking
Simulating data to gain insights into power and p-hacking
 
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
Ludäscher et al. 2014 - A Hybrid Diagnosis Approach Combining Black-Box and W...
 
Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401
 

Similar to Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test for AI

Artificial intelligence cs607 handouts lecture 11 - 45
Artificial intelligence   cs607 handouts lecture 11 - 45Artificial intelligence   cs607 handouts lecture 11 - 45
Artificial intelligence cs607 handouts lecture 11 - 45
Sattar kayani
 
artficial intelligence
artficial intelligenceartficial intelligence
artficial intelligence
Phanindra Mortha
 
Introduction to machine learning-2023-IT-AI and DS.pdf
Introduction to machine learning-2023-IT-AI and DS.pdfIntroduction to machine learning-2023-IT-AI and DS.pdf
Introduction to machine learning-2023-IT-AI and DS.pdf
SisayNegash4
 
Unit-V Machine Learning.ppt
Unit-V Machine Learning.pptUnit-V Machine Learning.ppt
Unit-V Machine Learning.ppt
Sharpmark256
 
Evolutionary deep learning: computer vision.
Evolutionary deep learning: computer vision.Evolutionary deep learning: computer vision.
Evolutionary deep learning: computer vision.
Olivier Teytaud
 
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
NUS Institute of Applied Learning Sciences and Educational Technology
 
Machine_Learning.pptx
Machine_Learning.pptxMachine_Learning.pptx
Machine_Learning.pptx
shubhamatak136
 
notes as .ppt
notes as .pptnotes as .ppt
notes as .pptbutest
 
Intro to machine learning
Intro to machine learningIntro to machine learning
Intro to machine learning
Akshay Kanchan
 
Introduction to Machine Learning.
Introduction to Machine Learning.Introduction to Machine Learning.
Introduction to Machine Learning.butest
 
Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401butest
 
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
NUS Institute of Applied Learning Sciences and Educational Technology
 
Reinforcement Learning
Reinforcement LearningReinforcement Learning
Reinforcement Learningbutest
 
Dowhy: An end-to-end library for causal inference
Dowhy: An end-to-end library for causal inferenceDowhy: An end-to-end library for causal inference
Dowhy: An end-to-end library for causal inference
Amit Sharma
 
F0422052058
F0422052058F0422052058
F0422052058
ijceronline
 

Similar to Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test for AI (20)

Artificial intelligence cs607 handouts lecture 11 - 45
Artificial intelligence   cs607 handouts lecture 11 - 45Artificial intelligence   cs607 handouts lecture 11 - 45
Artificial intelligence cs607 handouts lecture 11 - 45
 
S10
S10S10
S10
 
S10
S10S10
S10
 
artficial intelligence
artficial intelligenceartficial intelligence
artficial intelligence
 
Introduction to machine learning-2023-IT-AI and DS.pdf
Introduction to machine learning-2023-IT-AI and DS.pdfIntroduction to machine learning-2023-IT-AI and DS.pdf
Introduction to machine learning-2023-IT-AI and DS.pdf
 
Unit-V Machine Learning.ppt
Unit-V Machine Learning.pptUnit-V Machine Learning.ppt
Unit-V Machine Learning.ppt
 
ppt
pptppt
ppt
 
ppt
pptppt
ppt
 
Evolutionary deep learning: computer vision.
Evolutionary deep learning: computer vision.Evolutionary deep learning: computer vision.
Evolutionary deep learning: computer vision.
 
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
Joseph Jay Williams - WESST - Bridging Research and Practice via MOOClets & C...
 
Machine_Learning.pptx
Machine_Learning.pptxMachine_Learning.pptx
Machine_Learning.pptx
 
notes as .ppt
notes as .pptnotes as .ppt
notes as .ppt
 
Intro to machine learning
Intro to machine learningIntro to machine learning
Intro to machine learning
 
Introduction to Machine Learning.
Introduction to Machine Learning.Introduction to Machine Learning.
Introduction to Machine Learning.
 
151028_abajpai1
151028_abajpai1151028_abajpai1
151028_abajpai1
 
Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401Machine Learning: Foundations Course Number 0368403401
Machine Learning: Foundations Course Number 0368403401
 
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
Joseph Jay Williams - WESST - Bridging Research via MOOClets and Collaborativ...
 
Reinforcement Learning
Reinforcement LearningReinforcement Learning
Reinforcement Learning
 
Dowhy: An end-to-end library for causal inference
Dowhy: An end-to-end library for causal inferenceDowhy: An end-to-end library for causal inference
Dowhy: An end-to-end library for causal inference
 
F0422052058
F0422052058F0422052058
F0422052058
 

More from Codiax

Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Codiax
 
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluationCostas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Codiax
 
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Codiax
 
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Codiax
 
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Codiax
 
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Codiax
 
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videosAdria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Codiax
 
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
Codiax
 
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Codiax
 
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Codiax
 
Matthias Feys (ML6) – Bias in ML: A Technical Intro
Matthias Feys (ML6) – Bias in ML: A Technical IntroMatthias Feys (ML6) – Bias in ML: A Technical Intro
Matthias Feys (ML6) – Bias in ML: A Technical Intro
Codiax
 
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Codiax
 
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Codiax
 
Maciej Marek (Philip Morris International) - The Tools of The Trade
Maciej Marek (Philip Morris International) - The Tools of The TradeMaciej Marek (Philip Morris International) - The Tools of The Trade
Maciej Marek (Philip Morris International) - The Tools of The Trade
Codiax
 
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Codiax
 
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Codiax
 
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected WorldJakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Codiax
 
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Codiax
 
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Codiax
 
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network ServerAlexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Codiax
 

More from Codiax (20)

Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
 
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluationCostas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
 
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
 
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
 
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
 
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
 
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videosAdria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos
 
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Sat...
 
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
 
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
 
Matthias Feys (ML6) – Bias in ML: A Technical Intro
Matthias Feys (ML6) – Bias in ML: A Technical IntroMatthias Feys (ML6) – Bias in ML: A Technical Intro
Matthias Feys (ML6) – Bias in ML: A Technical Intro
 
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
 
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
 
Maciej Marek (Philip Morris International) - The Tools of The Trade
Maciej Marek (Philip Morris International) - The Tools of The TradeMaciej Marek (Philip Morris International) - The Tools of The Trade
Maciej Marek (Philip Morris International) - The Tools of The Trade
 
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
 
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
 
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected WorldJakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
 
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
 
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
 
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network ServerAlexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
 

Recently uploaded

Climate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing DaysClimate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing Days
Kari Kakkonen
 
DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
Kari Kakkonen
 
Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !
KatiaHIMEUR1
 
By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
Pierluigi Pugliese
 
Removing Uninteresting Bytes in Software Fuzzing
Removing Uninteresting Bytes in Software FuzzingRemoving Uninteresting Bytes in Software Fuzzing
Removing Uninteresting Bytes in Software Fuzzing
Aftab Hussain
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
Kumud Singh
 
GridMate - End to end testing is a critical piece to ensure quality and avoid...
GridMate - End to end testing is a critical piece to ensure quality and avoid...GridMate - End to end testing is a critical piece to ensure quality and avoid...
GridMate - End to end testing is a critical piece to ensure quality and avoid...
ThomasParaiso2
 
UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5
DianaGray10
 
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AIEnchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Vladimir Iglovikov, Ph.D.
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
sonjaschweigert1
 
20240605 QFM017 Machine Intelligence Reading List May 2024
20240605 QFM017 Machine Intelligence Reading List May 202420240605 QFM017 Machine Intelligence Reading List May 2024
20240605 QFM017 Machine Intelligence Reading List May 2024
Matthew Sinclair
 
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
James Anderson
 
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
SOFTTECHHUB
 
National Security Agency - NSA mobile device best practices
National Security Agency - NSA mobile device best practicesNational Security Agency - NSA mobile device best practices
National Security Agency - NSA mobile device best practices
Quotidiano Piemontese
 
Full-RAG: A modern architecture for hyper-personalization
Full-RAG: A modern architecture for hyper-personalizationFull-RAG: A modern architecture for hyper-personalization
Full-RAG: A modern architecture for hyper-personalization
Zilliz
 
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Nexer Digital
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Malak Abu Hammad
 
Microsoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdfMicrosoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdf
Uni Systems S.M.S.A.
 
RESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for studentsRESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for students
KAMESHS29
 
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
SOFTTECHHUB
 

Recently uploaded (20)

Climate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing DaysClimate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing Days
 
DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
 
Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !
 
By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
 
Removing Uninteresting Bytes in Software Fuzzing
Removing Uninteresting Bytes in Software FuzzingRemoving Uninteresting Bytes in Software Fuzzing
Removing Uninteresting Bytes in Software Fuzzing
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
 
GridMate - End to end testing is a critical piece to ensure quality and avoid...
GridMate - End to end testing is a critical piece to ensure quality and avoid...GridMate - End to end testing is a critical piece to ensure quality and avoid...
GridMate - End to end testing is a critical piece to ensure quality and avoid...
 
UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5UiPath Test Automation using UiPath Test Suite series, part 5
UiPath Test Automation using UiPath Test Suite series, part 5
 
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AIEnchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
 
20240605 QFM017 Machine Intelligence Reading List May 2024
20240605 QFM017 Machine Intelligence Reading List May 202420240605 QFM017 Machine Intelligence Reading List May 2024
20240605 QFM017 Machine Intelligence Reading List May 2024
 
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
 
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...
 
National Security Agency - NSA mobile device best practices
National Security Agency - NSA mobile device best practicesNational Security Agency - NSA mobile device best practices
National Security Agency - NSA mobile device best practices
 
Full-RAG: A modern architecture for hyper-personalization
Full-RAG: A modern architecture for hyper-personalizationFull-RAG: A modern architecture for hyper-personalization
Full-RAG: A modern architecture for hyper-personalization
 
Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?Elizabeth Buie - Older adults: Are we really designing for our future selves?
Elizabeth Buie - Older adults: Are we really designing for our future selves?
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
 
Microsoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdfMicrosoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdf
 
RESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for studentsRESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for students
 
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
Goodbye Windows 11: Make Way for Nitrux Linux 3.5.0!
 

Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test for AI

  • 1. Proving theorems: still a major test for AI Dr Sean Holden University of Cambridge Department of Computer Science and Technology and Trinity College Codiax 2019
  • 2. How did I get here? Machine learning research since 1989: • Initially on computational learning theory, particularly sample complexity of the cross-validation and other estimates. • Applications in drug design with Glaxo-SmithKline and others. • Organelle proteomics in collaboration with the Cambridge Department of Bio- chemistry. • Various other areas in Bayesian inference, survey propagation, protein interac- tions, gene expression and so on. Most recently: I’ve been working on the question of…
  • 3. How did I get here? How can machine learning be applied to proving mathematical theorems?
  • 4. How did I get here? Principia Mathematica… Logicism: is all mathematics just logic? • Alfred North Whitehead and Bertrand Russell attempted to demonstrate this. • Three volumes, published in 1910, 1912 and 1913. • Initially, very basic logical axioms. Then a systematic development. • Proofs in their system can easily be searched for by an automated theorem prover. Famously, the important theorem 1 + 1 = 2 appears on page 83 of Volume 2. Take-home message: being precise about mathematics is hard! However, this has formed a fundamental challenge in AI since its earliest days: the General Problem Solver, Herbert A. Simon, Cliff Shaw and Allen Newell, 1959.
  • 5. Proofs in propositional/Boolean logic Propositional logic: • Variables can be true (T) or false (F). • I have connectives ∧ (AND), ∨ (OR) and ¬ (NOT). 1. ¬x is T when x is F, and F when x is T. 2. x1 ∧ x2 is T precisely when both x1 and x2 are T. 3. x1 ∨ x2 is T precisely when x1 or x2 (or both) are T. • For example, the formula (x1 ∨ x2) ∧ x3 is T when x1 = T, x2 = F and x3 = T. Here is the game: I give you a HUGE formula with n variables: (x1 ∨ x2 ∨ ¬x3) ∧ (¬x2 ∨ · · · ∨ x10) ∧ · · · ∧ (¬x1,000,000) and you have to tell me whether or not there is a way of setting the variables to make the formula T. This is the satisfiability (SAT) problem. There is no known method that can guarantee to be better than testing all 2n possibilities!
  • 6. Proofs in first-order logic Am I a WEIRDO? ∀x . Pierced(x) ∧ Male(x) −→ Weirdo(x) Pierced(sean) Male(sean) Does Weirdo(sean) follow? We should probably agree that it does. (Hooray!)
  • 7. First-order logic The reason this is hard is simple: {W(sean)} {} {P(sean)} {M(sean)} x = sean The set of clauses grows There is a choice of which pair of clauses to resolve {¬W(sean)}{¬P(x), ¬M(x), W(x)} {¬M(sean), W(sean)} And it’s worse than it seems: • The full problem is semidecidable. • (You can prove a valid theorem, but you might not be able to disprove an invalid one.) • The propositional subproblem is NP-complete.
  • 8. Have computers managed to prove anything? The Robbins conjecture So automated theorem proving is known to be hard. But has it been used to prove any original mathematics? Robbins algebras: • Commutative: ∀a, b . a ∨ b = b ∨ a • Associative: ∀a, b, c . (a ∨ b) ∨ c = a ∨ (b ∨ c) • Robbins axiom ∀a, b . ¬(¬(a ∨ b) ∨ ¬(a ∨ ¬b)) = a Robbins conjecture: All Robbins algebras are Boolean. The proof was found by William McClune using the EQP theorem prover. (Jour- nal of Automated Reasoning, 1997.)
  • 9. So what’s the problem? So, what’s the problem? Although the Robbins conjecture represents a proof of a genuine theorem by an automated theorem prover: • It is something of an isolated result. • It essentially represents an application of brute-force search. • The proof has little if any resemblance to the kind of proof a person might construct. • It’s not enough to impress the mathematicians. What will it take to impress them? We could aim to prove something BIG! We could aim to prove multiple results with the same system without human guidance. (Domain expertise used to tune the prover.)
  • 10. First-order logic + machine learning There’s a growing interest in improving ATPs by adding machine learning to them. I’ve been interested in this for some time: • The E theorem prover is a state-of-the-art prover for predicate logic. • One way it tries to address the problem is to use features of the theorem of interest to choose from 82 heuristics for choosing what to resolve. • Can we improve on this? We found that, yes, we could. (Bridge, Holden and Paulson, Journal of Automated Reasoning, 2014). Specifically, applying support vector machines and Gaussian process classifiers we achieved comparable performace with only 5 heuristics. Take-home message: machine learning can provide useful guidance for automated theorem provers.
  • 11. This proof is quite big… The classification theorem for finite simple groups Every finite, simple group is isomorphic to one of: • a cyclic group with prime order • an alternating group with degree ≥ 5 • a group of Lie type • one of 26 sporadic groups.
  • 12. This proof is quite big… The proof of the theorem was announced in the early 1980s. BUT it was, er, quite big: • Estimated at > 10, 000 pages long. • After announcement, it turned out there was a gap related to quasithin groups… • … which took another 7 years and 1221 pages. • There is an effort at present to shorten the proof . • As of 2019 this is 8 volumes, with another 5 likely to be needed. QUESTION: Is this proof correct? (The mathematicians don’t seem too worried, but I’m not so sure…)
  • 13. Proof checking: formalising mathematical results Systems such as Isabelle provide a means of constructing formally verified proofs: https://www.cl.cam.ac.uk/research/hvg/Isabelle/ Such systems are often presented as proof assistants. There is a body of formalized mathematics: the Archive of Formal Proofs (AFP): https://www.isa-afp.org/ This provides us with another target for machine learning: Lemma recommendation
  • 14. Machine learning for selecting premises Current work on lemma recommendation (Słowik, Mangla, Jamnik, Holden, Paul- son, AAAI 2019). -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x -2 -1 0 1 2 3 4 5 f(x) Squared exponential prior, l = 1/2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x -3 -2 -1 0 1 2 3 4 5 f(x) γ-exponential prior, l = γ = 1/2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x -2 -1 0 1 2 3 4 5 6f(x) Rational quadratic prior, l = a = 3 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x -2 -1 0 1 2 3 4 5 f(x) Neural network prior, σ1 = 15, σ2 = 10 • We build on the existing Sumo Inference Engine (SInE). • We use Bayesian optimization to tune SInE. • The AFP is a natural source of training data.
  • 15. Mathematical language is tricky Question: why don’t more mathematicians use proof assistants? Answer: mathematicians don’t necessarily want to learn a new formal language. CHALLENGE Is there a way of writing mathematics that looks natural while being formally verifiable? It turns out that this is REALLY HARD! See for example The Language of Mathematics: A Linguistic and Philosophical Investigation, by Mohan Ganesalingam, Springer, 2013.