King of AI.. Time Magazine & Arvind Krishna trying to reclaim it etc.... Also Yoshua Bengio, Yann LeCun and Geoffrey Hinton

Hey Readers, 

It's 4 AM..and I am blogging again. This article is stuck in my head by Times Magazine where Arvind Krishna mentioned about reclaiming the title of King of AI ( https://time.com/6167753/arvind-krishna-ibm-ceo-interview/ ). 

Believe me, I absolutely like Satya, Sundar & Arvind Krishna's leadership style (while Alteryx IPO has been successful, Dataiku & SAS are going for an IPO before Jan 2024, Alibaba Cloud continues to focus on AliGraph while Sundar Pichai talks about Edge computing & federated learning in his latest interview starting at 27:50 min - https://www.youtube.com/watch?v=EuF8nv53JeI&t=1997s). Very pragmatic in their approach. But, in all the honesty & humble opinion, the king of AI will be the one who is

1. Able to bring paradigm shift in the market with new AI methods.. and probably bridge the gap between Research & Industry 

2. Is able to differentiate role of Responsible AI, Narrow AI, True AI & so on...

3. Bring continuous thought leadership in use cases (e.g. Amazon's Astra) 

4. Drive adoption not only in big businesses with million $ accounts but small businesses, start-ups, local banks, local businesses, small clinics where AI & data has never been thought about.. 

5. The one that governs the data & AI world like never before with wisdom & unique ability to navigate through any domain with flawless ability to invent & reinvent new methods continuously & industrialize older methods for optimum top-line & bottom-line while seamlessly talking about other technologies including AR/VR, Blockchain, Hardware, Chip Manufacturing and so on.. 

6. Who is also able to navigate through physics, quantum, Economics, Social Sciences fledgling branches like Econophysics, SocioPhysics, Swarm Intelligence, Cobotics etc. since AI is such an interdisciplinary field which is continuous evolving. 

7. Who is also able to understand the history of AI which is put together in many books including Genius Makers and so on.. 

One guy/org to rule them all at the same time.. Pretty fancy definition right?... 

Anyway coming back to 2018 debate/discussions between Yoshua Bengio, Yann LeCun and Geoffrey Hinton which is covered in Kdnuggets - https://www.kdnuggets.com/2017/10/rework-interview-yoshua-bengio-yann-lecun-geoffrey-hinton.html  

3 highlights -

1. Bengio used to call Geoff - Probability Police :) Fancy name 

2. When asked about what advice you’d share with young people working in AI now - Bengio says 'Listen to Geoff' Pretty Fancy answer..

3. Lecun talks about Bias & partnership in AI 

Adelyn Zhou does an amazing job of conducting an interview.. 

 

 

Maryam Mirzakhani & Forbes, Teichmüller Space, WeChat - Fully Hyperbolic Neural Networks

Hi Readers,

I had a super-busy day at work yesterday and wrote a blog at 1 AM. But, today I have taken a leave to spend some time to chill out.  

--- One of the reasons I speak about Mirror Symmetry and Maryam Mirzakhani is this Forbes article- https://www.forbes.com/sites/startswithabang/2017/08/01/maryam-mirzakhani-a-candle-illuminating-the-dark/?sh=5aa7449836c1 

Sometimes Forbes writes really cool articles right.. I hope they get right influencers to spread awareness of such brilliant articles through influencer marketing while understanding stubborn & non-stubborn players in social networks with opinion mining - Ref - https://arxiv.org/abs/1609.03465 (it would be cool & optimum to spread awareness of math articles through math based influencer marketing right?)  

--- One of the brilliant papers by Maryam Mirzakhani is Volume Growth on Teichmuller Space which speaks about topological entropy of the geodesic flow. (September 20, 2011- Lattice point asymptotics and volume growth on Teichmüller space Jayadev Athreya, Alexander Bufetov, Alex Eskin, Maryam Mirzakhani) Kind of cool paper to study Geodesic CNNs (Geodesic Convolutional Neural Networks bit more)...  

--- The paper Hyperbolic spaces in Teichmüller Space by Christopher J. Leininger and Saul Schleimer explains the way hyperbolic space almost-isometrically embeds into the Teichmüller Space (  https://arxiv.org/abs/1110.6526 ) kind of very smart inspiration from Mirzakhani's paper.. 

--- which is further adopted in Hyperbolic Neural Networks by Octavian-Eugen Ganea ( https://arxiv.org/abs/1805.09112 )which focuses on hyperbolic embeddings - Möbius gyrovector spaces with the Riemannian geometry of the Poincaré model of hyperbolic spaces which helps to embed sequential data and perform classification in the hyperbolic space (for further Poincare Glove as well - HyperE at Stanford)  

While this paper cites Sanja Fidler's paper on Order Embeddings, One of the best papers which has recently come out is by WeChat - Fully Hyperbolic Neural Networks .. one of the smartest point in the design being.. Fully Hyperbolic Attention Layer - https://arxiv.org/pdf/2105.14686.pdf

1.Yang-Mills, 2.Job Postings 3.Calabi-Yau 4.EconoPhysics, Divergent Recovery, TopologyGAN, Ricci Flow, Hodge theory

Hi Readers,

Yang Mills - First of all, lot of people must have seen my old orchestration on Yang Mills & Social Collider - Kdnuggets (and Neural Networks & Ricci Flow my old blog) - Here comes the absolutely brilliant paper on Yang Mills almost depicting the same - Emergent Attention on Yang-Mills Space of Connections - https://lukepereira.github.io/notebooks/documents/2021-moduli-attention/main.pdf (from Universidade Federal Fluminense .... Brazil & India is where so much of mathematical innovation is happening right ) Btw, you should watch my Ricci Flow video on Community detection & clustering to before reading this paper - https://www.youtube.com/watch?v=z_pjsJisdHQ 

Job Postings - This is an absolutely brilliant paper by Pratik Kothari on
Job Postings and Aggregate Stock Returns A much needed paper in today's time especially when CNBC announced Stock futures bounce as investors assess start of new quarter, bond market recession ( https://www.cnbc.com/2022/03/31/stock-market-futures-open-to-close-news.html ) 
- https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3223141

Calabi Yau - While I am impressed with the Inception Neural Network for Complete Intersection Calabi-Yau 3-folds - inspired by Google’s Inception model to compute the Hodge number I am still waiting for application on DeepSwarm in Calabi Yau. 

EconoPhysics - Dr. Gita Gopinath mentions Divergent Recovery in her IMF blog. How to make it optimum can be a very cool question to solve through EconoPhysics?

Most of you might have read my previous blog on Ricci Curvature & Market Failure ( https://researchcircle.blogspot.com/2022/02/job-posting-paper-ricci-flow-economics.html ). And some of you might have seen my recent tweet on Hodge Theory & Trading Networks in the evening. ( https://twitter.com/Prasad_Kothari/status/1509634908125335559 )

The following figure is from Sandhu's paper (w.r.t. Average Ricci curvature over a 15-year span of the S&P 500 - Choosing a window of T = 22 days )

I have mentioned portrait divergence in the following orchestration to compare sub-topologies between the economic & trade networks.  

Reference - 

1. Symmetry perception with spiking neural networks -  https://www.nature.com/articles/s41598-021-85232-3

2. Comparing methods for comparing networks -  https://www.nature.com/articles/s41598-019-53708-y

3. Ricci Curvature-Based Semi-Supervised Learning on an Attributed Network - https://ui.adsabs.harvard.edu/abs/2021Entrp..23..292W/abstract

4. Controlling Entropy via Discrete Ricci Flow Over Networks - https://arxiv.org/abs/1910.04560

The most interesting thing which I have not mentioned in this solution (apart from the code) is Thermodynamic Entropy in Quantum Statistics for Stock Market Networks from Shanghai University which talks about - Asian Financial Crisis (what could have been cool is use of HodgeNet (Graph Neural Networks for Edge Data) in it - https://ieeexplore.ieee.org/document/9049000

Edge entropy distribution of network structure before, during, and after the 1997 Asian financial crisis. (a)-(e) Bose-Einstein statistics. (d)-(f) Fermi-Dirac statistics.

  

Mirror Symmetry, Swarm Intelligence, Calibi Yau, GAN, SNNs, Econophysics

Hey Readers,

I am back again with a new blogpost on Saturday morning 5:30 AM. I got lot of inquiries about Dennis Sullivan and his amazing victory on Abel Prize. One of the best papers I have read by Dennis Sullivan is Circle Packing (off course my NLP, Reinforcement Learning, Ant Colony Optimization, Profiling etc. based prediction on last 50 years of math research papers is taking shape with Bhargav Bhatt's Hodge Theory, Maryna Viazovska's sphere packing etc. while off course Terence Tao has just published another paper on Sphere packing.. Sphere packing seems trend of the year while I am already thinking Duck pack algorithm — A new swarm intelligence algorithm for route planning based on imprinting behavior).. 

I also used to write a lot about Econophysics, String Theory (M-Theory etc. as I was big fan of Stephen Hawkings and probably had entire Brief History of Time by heart that time) Hodge Theory, Hodge Decomposition of information flow in networks & its impact on markets. 

Perelman's 2008 paper The entropy formula for the Ricci flow and its geometric applications says that he did not know enough about Blackholes & (pseudo)-riemannian geometry, Ricci Flow, Black Hole Thermodynamics, etc https://arxiv.org/pdf/math/0211159.pdf

Hence, I thought a perfect way to apply Perelman's work was AI, Data Science (please see my Ricci Flow- https://www.youtube.com/watch?v=z_pjsJisdHQ&t=26s ) & Econophysics using black hole thermodynamics etc. 

I see few papers published on this topic these days -

1. Trading networks and Hodge theory - https://iopscience.iop.org/article/10.1088/2399-6528/abd1c2/pdf

2. Semiparametric Estimates of Monetary Policy Effects: String Theory Revisited - https://economics.mit.edu/files/13030

3. Ricci Solitons & Black Holes - http://dspace.rri.res.in/bitstream/2289/3857/2/Chapter_7.pdf

4. Entropy of Reissner–Nordström 3D Black Hole in Roegenian Economics - https://www.semanticscholar.org/paper/Entropy-of-Reissner%E2%80%93Nordstr%C3%B6m-3D-Black-Hole-in-Udriste-Ferrara/fb2052f27e6700fed05ae259d63232639b465643

 

I used to write a lot about swarm intelligence (ant colony optimization etc.) on Google Plus, G-spaces long time back on this blog to give mathematical meaning to Google Plus (Hodge Cycle, Supersymmetric Yang Mills, Social Collider etc.) & also give some movie meaning to it. Also, I wrote a lot about Ricci Flow (Used to prove Poincare Conjecture) -

1. Meet the parents, Circle of Trust and Google Plus - https://www.kdnuggets.com/2013/10/circle-trust-google-plus.html

2. Yang-Mills: A million dollar connection between Twitter and quantum physics - https://www.kdnuggets.com/2013/11/yang-mills-million-dollar-connection-twitter-quantum-physics.html

3. A lot about Ricci Flow, Group theory, Quantum Groups (remember Sir Atiyah's 1984 paper with TIFR which I cited few blogposts back)

4. I also wrote a lot about applications of deep learning & swarm intelligence to solve mathematical conjectures 

Hence the new orchestration which Cao & Zhao might like (as I see their paper on Calabi Yau, Machine Learning and Convolutional Neural Networks .. I would like to see possible use of Geodesic Convolutional Neural Networks) on Yang Mills which also uses Maryam Mirzakhani's concept on Mirror Symmetry in Spiking Neural Networks... 

SYZ conjecture, Maryam Mirzakhani - AI, Geodesics, Riemannian surfaces, Dennis Sullivan

Hey readers,

Everyone must have seen my last comment on Andrica's conjecture & Firoozbakht's conjecture and how looking at prime numbers as dynamical system is a smarter way to solve this. I also have a way to look at it further in terms of How to use GANs - Generative Adversarial Networks (with optimum convergence while leveraging Nash entropy) to solve PDEs in Dynamical & Chaotic systems generated by prime numbers.

There was this beautiful paper from Maryam Mirzakhani (field prize winner 2014) - Simple geodesics and Weil-Petersson volumes of moduli spaces of bordered Riemann surfaces (2005) - https://www.math.stonybrook.edu/~mlyubich/Archive/Geometry/Teichmuller%20Space/Mirz3.pdf

A lot of my discussions during my evening time in 2010 happened on how to apply her paper in AI. So, I came up with an orchestration in 2011 on Geodesics CNN on Riemannian Surface (I used to write a same blog by the name of DataHulk - which I deleted in 2017 and restarted this year).

Here is a very interesting paper which got published in Switzerland on the same topic - Geodesic convolutional neural networks on Riemannian manifolds - https://arxiv.org/pdf/1501.06297.pdf . The most interesting part of this paper was heat diffusion on manifolds, Heat kernel signature etc.

This was followed by my 2 lectures & video on

1. Ricci Flow (heat diffusion equation with derivative form) in Networks ( https://www.kdnuggets.com/2014/05/poincare-conjecture-perelman-topology-social-networks.html )

2. Some part of my lecture in SDSU which focused on Pruning of Neural Networks using Ricci Flow after which University of Cambridge published a new paper on 
RicciNets: Curvature-guided Pruning of High-performance Neural Networks Using Ricci Flow - https://arxiv.org/abs/2007.04216

3. You might like my video on Ricci Flow & its use in use in community detection  - https://www.youtube.com/watch?v=z_pjsJisdHQ&t=23s  

Here is one way to look at SYZ conjecture -

By the way you might like the old paper - THE HOMOLOGY THEORY OF THE CLOSED GEODESIC PROBLEM MICHELINE VIGUE-POIRRIER & DENNIS SULLIVAN - https://www.math.stonybrook.edu/~dennis/publications/PDF/DS-pub-0034.pdf  ( Dennis Sullivan won Abel prize on 24th March for amazing circle packing work )

Edited - Zhiren Wang, associate professor of mathematics at Penn State, has been award the 11th Brin Prize in Dynamical Systems MAY 24, 2022 

PK the genius, Misinformation detection in networks including Vimeo, TopologyGANs, HodgeNet, Duck Swarm intelligence

Hey readers, 

Happy Sunday morning! A lot of my readers liked my blogpost on Econophysics, Ricci Flow and Gita Gopinath's post on Divergent Recovery ( https://researchcircle.blogspot.com/2022/02/ricci-flow-geomstats-lie-groups-on.html ) 

Also, Lot of my readers liked my video on Ricci Flow Animation - 

https://www.youtube.com/watch?v=z_pjsJisdHQ&t=16s 

Also lot of people liked my previous blog about Duck Swarm Intelligence, Ricci Flow (my tweet few months back on TopologyGANs & HodgeNet).. 

Few years back I met Raghuram Rajan at Chicago Booth to discuss his book & I also told him that I am working on new & innovative approach to detect misinformation over networks (I know there are lot of approaches but nothing like this one & yes this is different from the one I wrote about Inorganic Social Network Growth through Ant Colony Optimization in my old DataHullk - researchcircles blog.. I will write more about it in my next one). 

Every social network including Vimeo faces a problem of Misinformation. Here is my new orchestration to detect misinformation in social networks (I have mentioned about Hodge Decomposition of information Flow over narrow or complex networks) 

- 

Firoozbakht's conjecture, BDS conjecture & deep learning, Lie Groups, Statistical Riemannian Framework, Ramanujan Machine..

Hey Readers,

It's 4:30 AM Saturday night. And I am at it again. A blog I restarted which I used to write 12 years back to motivate new thinking in data, math and AI. The paper I like the most these days is Job Postings and Aggregate Stock Returns - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3223141

1. If I ever write my next book, there will be special mention of Iranian mathematician Farideh Firoozbakht - almost an unsung hero of math - She studied pharmacology and later mathematics at the University of Isfahan, and later taught mathematics at that university (Talk about reinventing the career). 

Extremely famous for her Firoozbakht's conjecture (pattern in prime numbers in simple words) which was stated in 1982 - this conjecture achieved some fame 

- in 2011 when I wrote about it along with new orchestrations on dynamical systems (Naive Entropy, Nash Entropy etc), 

- also on Maryam Mirzakhani's work on topology, geometry and dynamical systems (specially her work on Simple geodesics and Weil-Petersson volumes of moduli spaces of bordered Riemann surfaces - https://www.math.stonybrook.edu/~mlyubich/Archive/Geometry/Teichmuller%20Space/Mirz3.pdf And when I also wrote about possibility of GeoDesic Convolutional Neural Networks - Geodesic CNNs & 

- & on Laura De Marco's work on dynamical systems my datahulk blog & 

in 2015 when Alexei Kourbatov published a paper on it in International Mathematical Forum 10 (2015), 283–288. Verification of the Firoozbakht conjecture for primes up to four quintillion, (unproven till now and has a very strong affiliation with Riemann Hypothesis).

I have a very strong viewpoint that Andrica's conjecture & Firoozbakht's conjecture and how looking at prime numbers as a dynamical system can have a smarter way to solve this. I also have a way to look at it further in terms of How To use GANs - Generative Adversarial Networks (with optimum convergence while leveraging Nash entropy) to solve PDEs in Dynamical & Chaotic systems generated through prime numbers. Anyway, more to come later... some surprises in store along with my Ricci Flow Video..  

Firoozbakht passed away in 2019 without many people knowing it... Almost an unsung hero who could have been celebrated everywhere. I always found it bit disappointing that Terence Tao mentioned Cramér conjecture in his blog but not Firoozbakht's conjecture. But, hey, everyone makes their own choice (that's why Thaler's choice architecture right??? right????) 

2. Another person who is doing some cool work is Nina Miolane - Amazing work on Statistics on Lie groups (& abelian groups): A need to go beyond the pseudo-Riemannian framework - https://aip.scitation.org/doi/abs/10.1063/1.4905963

Again, I want to highlight 1984 paper by Sir Atiyah, TIFR and Sir Atiyah's Student George Lusztig and his work on Quantum Groups. I know Nina's GeomStats is focus on Biomedical Topology but there is one more way to grow GeomStats in topology

i.e. growing GeomStats like Ramanujan Machine where AI would solve complex group theory problems trained on Lusztig's group theory paper including lie groups, abelian groups, Weyl groups, Chevalley groups, semisimple p-adic groups, quantum groups etc..

Btw, The first suggestion I gave to Ramanujan Machine when I became their advisor was this paper on BDS conjecture & deep learning by Laura Alessandretti - Machine Learning meets Number Theory: The Data Science of Birch-Swinnerton-Dyer - https://arxiv.org/abs/1911.02008

Security, Topological Game Theory & Ricci Flow

Hey guys,

Thanks again for around 40,000 views of my last post. It's pretty cool to get these many views for a nerdy blog. And apologies, I have been super-busy at work, writing books, research papers, articles, giving guest lectures etc.

This post is for Perelman, Dr Yau, Melanie Weber, Cao and Zhu, Dr. Fefferman & all nerds who understand my Ricci Flow Video on youtube.. This is my entry into the Field prize Math Video Contest ..win win win.. - https://www.youtube.com/watch?v=z_pjsJisdHQ&t=6s  )...

Here you go... 

1. Amazing Amazing book by Milind Tambe on Security and Game Theory: Algorithms, Deployed Systems - https://www.amazon.com/Security-Game-Theory-Algorithms-Deployed/dp/1107096421 

The book provides detailed overview on

1. Game theoretical framework for security measures,
2. Intelligent Randomization over routes,
3. Bayesian Stackelberg Games,
4. Robust Game Theory,
5. Bayesian Nash Equilibrium,
6. Strategic Security Allocation in Transport networks,
7. Randomization with partial adversary model,
8. Stackelberg vs Nash in security games

2. I want to push this book and concept further in terms of Ricci Flow, Evolutionary Game theory & Security measures - 

Evolutionary game theory is a continuous model with interactions (frequency dependent selection). e.g. the classical Lotka-Volterra predator-prey system fits into this framework. Another example is an infinite population of people playing the game rock-paper-scissors in continuous time. Obviously, if almost everyone is playing rock, the trend will be for more people to play paper. Asymptotic behavior, Nash equilibria and stability of fixed points are studied.

This framework may also be used to analyze the evolution of geometric structures. The Ricci flow is exactly a “replicator equation of quadratic type” for evolutionary game theory. New evolutionary models for various types of geometric flows are put forward.

Hence application of Evolutionary game theory, the classical Lotka-Volterra predator-prey system with The Ricci flow as exactly a “replicator equation of quadratic type” for evolutionary game theory for defense strategy. 

For Example - 

Defense Strategy Selection Model Based on Multistage Evolutionary Game Theory - https://www.hindawi.com/journals/scn/2021/4773894/

The existing network attack and defense analysis methods based on evolutionary games adopt the bounded rationality hypothesis. However, the existing research ignores that both sides of the game get more information about each other with the deepening of the network attack and defense game, which may cause the attacker to crack a certain type of defense strategy, resulting in an invalid defense strategy. The failure of the defense strategy reduces the accuracy and guidance value of existing methods.

3. Ricci Flow and Community Detection for defense and fraud detection. My video is self explanatory -

https://www.youtube.com/watch?v=z_pjsJisdHQ&t=6s 

The community detection application is not only limited to fraud detection but also for applications on biological networks, protein-protein networks, metabolic networks, and gene networks, etc. 

Ricci Flow, Geomstats, Lie Groups on Julia & EconoPhysics (and belated Happy Valentines day)

Hey Readers,

Again... thanks for reading my blog. I have been super-busy with lot of work. But,  30,000 views in single night is pretty cool. It is 9 PM in the evening, I just came from my walk in Maryland and I am on a mission to motivate everyone to learn more about data, AI, Math, Topology and so on. I should give more guest lectures when I get time.. but, for now please have a look at wonderful lecture by Geometric Statistics in Machine Learning GeomStats with Nina Miolane - https://www.youtube.com/watch?v=3yNInHuIWlY&t=1251s 

Hence, the next blogpost in the night on Ricci Flow (the topological concept that Grigori Perelman used to prove Poincare Conjecture) and Network Geometry.

1. Network geometry and market instability -https://royalsocietypublishing.org/doi/10.1098/rsos.201734

This paper talks about Time series of log-returns over a 32-year period (1985–2016) with network of stocks with 

1. Ollivier–Ricci (ORE), 
2. Forman–Ricci (FRE), 
3. Menger–Ricci (MRE), 
4. Haantjes–Ricci (HRE) 
5. Algebraic topological aspects, such as the homology groups and Betti numbers

minimum risk Markowitz portfolio of all the stocks, to better understand tipping points, systemic risk and resilience in financial networks, and enable us to develop monitoring tools required for the highly interconnected financial systems and perhaps forecast future financial crises and market slowdowns with use of Python package NetworkX (Yeah.. I used to use a lot of R, SAS, SQL, Cytoscape, Python in 2006 but Julia became my favorite language in 2012).

You can also check out few packages such as 

1. TheanoGeometry ( https://arxiv.org/abs/1712.08364 )- Riemann, Ricci and scalar curvature & geodesics  
2. Geomstats ( https://arxiv.org/abs/2004.04667 )   
3. McTorch  - McTorch, a manifold optimization library for deep learning ( https://arxiv.org/abs/1810.01811 ) 
4. Pymanopt: A Python Toolbox for Manifold Optimization using Automatic Differentiation ( https://arxiv.org/abs/1603.03236 )
5. Topological Entropy for Geodesic Flows under a Ricci Curvature condition: Jacobi field, Ricci curvature, topological entropy, tangent bundle, Negative Ricci curvature, isometry group and Injectivity radius - ( https://www.ams.org/journals/proc/1997-125-06/S0002-9939-97-03780-5/S0002-9939-97-03780-5.pdf )
6. And how can I forget my new favorite Julia for manifold Manifolds.jl: An Extensible Julia Framework for Data Analysis on Manifolds - a fast and easy to use library of Riemannian manifolds and Lie groups - ( https://arxiv.org/pdf/2106.08777.pdf ) 

Ricci Flow, Economics & a brilliant blog by Gita Gopinath

Hey readers,

It's 2:40 AM on Saturday night. Thanks for reading my DataHulk blog. I was glad to see 20,000 views in a single night on my last post - Yang Mills, Vector bundles, Quantum Information Geometry & Fisher-Bures Adversary Graph Convolutional Networks etc. And Yes, I noticed Terence Tao wrote his blog on the same day talking about Sphere Pinching paper (Smart people think alike I guess) - https://terrytao.wordpress.com/2022/02/08/perfectly-packing-a-square-by-squares-of-nearly-harmonic-sidelength/ 

I used to write this blog 12 years back to motivate everyone to learn math, data & AI folks. By the way, I have an exciting blogpost today which might change the way some of my followers look at economics (I know.. I know.. I promised to write about Spin Neural Networks, diffusion and capsule networks in my last blog.. which I will write later). 

1. Ricci Flow & Economics (And why Gita Gopinath's sentence in her blog is genius) - 

Here is an interesting paper followed by 2013-14 Ricci Flow & Poincare Conjecture in topology of Social Networks & Information Geometry article ( https://www.kdnuggets.com/2014/05/poincare-conjecture-perelman-topology-social-networks.html ).... Ricci curvature: An economic indicator for market fragility and systemic risk - https://www.researchgate.net/publication/303600815_Ricci_curvature_An_economic_indicator_for_market_fragility_and_systemic_risk which is an analysis based on geometric feature extraction on network data of daily returns from a set of stocks & market topology comprising the Standard and Poor’s 500 (S&P 500) over a 15-year span to highlight the fact that corresponding changes in Ricci curvature, is negatively correlated to increases in network fragility. 

network, is negatively correlated to increases in network fragility. To illustrate this insight, we examine daily

returns from a set of stocks comprising the Standard and Poor’s 500 (S&P 500) over a 15-year span to highlight

the fact that corresponding changes in Ricci curvature constitute a financial “crash hallmark.”

  

Average Ricci curvature over a 15-year span of the S&P 500 - Choosing a window of T = 22 days, we see that curvature captures several financial crashes and show that, on average, market behavior is fragile.

2. Gita Gopinath - 

Here is an interesting & brilliant blog (April 2021) by Gita Gopinath on divergent recovery -
Managing Divergent Recoveries - https://blogs.imf.org/2021/04/06/managing-divergent-recoveries/ speaking about 

1. Financial risks, 

2. Financial stability risks using macro-prudential tools, 

3. Cross-country gaps, global poverty reduction, 

4. Emerging markets, 

5. Divergent recovery paths, 

6. High degree of uncertainty and developing economies 

7. International liquidity

8. Liquidity protection

9. Debt restructuring 

10. financial stability risks using macro-prudential tools

11. Withdrawn loan payments, firm insolvencies

12. Cross-border profit shifting

which reminded me of network entropy, nodal entropy, geodesics curvature & divergence mentioned in the Sandhu's paper - Ricci curvature & Ricci Flow: An economic indicator for market fragility and systemic risk and my KDnuggets article on which intended to perform Ricci Flow on Social network analysis & information economics. 

Yang Mills, Vector bundles, Quantum Information Geometry & Fisher-Bures Adversary Graph Convolutional Networks etc.

Hello guys,


It's been a long time since my last post & its Tuesday night.. Yes.. I am still DataHulk.. When I get angry I work on Big Data... :) 

- And off course.. Maryna Viazovska (one of the researcher who might win field prize as per my first blogpost based on Graph Mining, NLP, Reinforcement Learning & Profiling along with Bhargav Bhatt etc. - Graph Mining, Network Science + Topology & Predicting Field Prize - https://researchcircle.blogspot.com/2021/12/graph-mining-network-science-topology.html) has published new paper on Hyperbolic Fourier series - https://arxiv.org/abs/2110.00148

....... while Terence Tao has published new paper on Perfectly packing a square by squares of nearly harmonic sidelength - https://arxiv.org/abs/2202.03594 (Another signal that Sphere Packing problem is flavor of the season and Maryna Viazovska might win Field Prize for her effort on Sphere Packing in N = 8 & 24) 

There are 4 papers which I suggest reading everyone who is interested in intersection of Yang Mills, Vector Bundles & Graph neural networks -

1. Fisher-Bures Adversary Graph Convolutional Networks (quantum information geometry & Fisher information of the neural network) -

http://proceedings.mlr.press/v115/sun20a.html

You can enhance this Graph Neural Network further with reconstructing Quantum geometry from Quantum Information from the following paper. 

2. Reconstructing Quantum Geometry from Quantum Information 

https://arxiv.org/pdf/gr-qc/0501075.pdf

2. The Yang-Mills α-flow in vector bundles over four manifolds and its applications (Gang Tian) - 

https://arxiv.org/abs/1303.0628

4. Vector Bundles On Algebraic Varieties (This is the most important paper by Dr. Atiyah & team from 1984 in collaboration with TIFR Mumbai & Oxford University) -

http://www.math.tifr.res.in/~publ/studies/Vector-Bundles-On-Algebraic-Varieties.pdf

One of these days, I will write in detail why Yang mills & vector bundles has potential solution while working on graph neural networks because of Quantum Information Geometry connection. It's going to be one hell of an orchestration. 

For now, look at the beauty of Sir Atiyah's 1984 paper in collaboration with TIFR. (BTW, Update 8 days after this blogpost was published.. on 17th Feb Sir Atiyah's Student George Lusztig won Wolf Prize for representation theory & Quantum Groups) 

My next blog is going to be about Spin Networks.    

Federated learning & Security (Story of a CEO & 2 researchers)

Hello Followers & Data Friends,

Again, thanks to someone who wrote this in my book review and helped me to restart this blog again.. I am glad someone still remembers my old blog and so many got motivated to enter AI & Data Science after reading it. My humble thank you - 

It's Sunday night (11:30 PM) and amazing times that we live in where AI and Data is the new oil. And it's also the time to have a deep look at Sundar's last month's interview where he spoke about Federated Learning (starting at 27:50). (Yeah, it was another busy weekend with lot of work but I still listen to Sundar's speeches on my walk) - https://www.youtube.com/watch?v=EuF8nv53JeI&t=1997s

Federated Learning & Edge Computing is an evolving topic most popular since Google announced its cookie-less marketing world last year - https://www.gartner.com/en/marketing/insights/articles/three-steps-for-marketers-to-prepare-for-a-cookieless-world

Sundar also spoke about Privacy, AI and a lot of futuristic things.

The books/papers that came to my mind immediately after listening to the interview are -

1. Milind Tambe's book on Security & Game Theory (2011) - https://teamcore.seas.harvard.edu/publications/security-and-game-theory-algorithms-deployed-systems-lessons-learned

This is one game theory book I liked a lot in 2011 and wrote about it in my blog that time. It was fresh with good perspective security & game theory while it was written in a simple language. Also, amazing amazing chapter 12 on Stackelberg versus Nash in Security Games. Every game theory & security student should read this chapter and if possible attend Milind Tambe's lectures. His lectures have amazing clarity & awesome vision. (Yes. I don't promote my own books 😜)

2. Advances and Open Problems in Federated Learning (2019) - https://arxiv.org/abs/1912.04977 by Ramesh Raskar & the team. A wonderful wonderful paper on Federated Learning. 

The chapter I Like in this paper/booklet is - Adapting ML Workflows for Federated Learning, Hyperparameter Tuning and Neural Architecture Search (NAS) which is inspired by Chaoyang He, Murali Annavaram, and Salman Avestimehr' s paper on FEDNAS - Federated deep learning via neural architecture search. 

I will try to cover these things in my any upcoming guest lectures very soon. I have been extremely busy. For now, it's time to listen to Milind Tambe's all lectures from the past including AI for Social Good lecture including his JPMorgan lecture... 

More coming soon on this topic...

Interesting case of Schrodinger equation & deep learning (and possible application in Commutative Hodge Conjecture)

If my readers remember my old blog (written under the pseudoname of DataHulk 😏), I used to talk about Physics Informed Neural Networks, Physics formed deep learning & Data, Machine learning & Deep learning to solve math conjectures 12 years back... much before The Ramanujan Machine (AI to Solve Math Conjectures) was born. 

And Yes, I saw review on my 2013 book - Healthcare Social Media Management and Analytics, so I know some of my readers have missed me 🙌🙌🙌





1. Here is a curious case which was written few months back - Data-driven vector soliton solutions of coupled nonlinear Schrödinger equation using a deep learning algorithm - https://www.researchgate.net/publication/355789067_Data-driven_vector_soliton_solutions_of_coupled_nonlinear_Schrodinger_equation_using_a_deep_learning_algorithm In this paper, there is pre-fixed multi-stage training algorithm by combining the error measurement & multi-stage training. This algorithm is much better suited for different dynamical behaviors of solitons with faster convergence rate.


2. Recently, G. Tabuada from MIT proposed a series of noncommutative counterparts all conjectures including Grothendieck standard conjecture, Voevodsky nilpotence conjecture, Tate conjecture, Weil conjecture etc. XUN LIN has also proposed NON-COMMUTATIVE HODGE CONJECTURE.


Similar physics informed neural nets or GANs can be used for commutative Hodge Conjecture. More to come soon on this topic.


On a different topic, Please do listen to András Juhász & Marc Lackenby (similar to University of Sydney mathematician Geordie Williamson's work with DeepMind on representation theory) - https://www.youtube.com/watch?v=hIUiPi-jAjM

Swarm Intelligence - Another curious case of Duck Swarm algorithm & New Google algorithm on Pathway Analysis

Hey guys, Thanks for reading my restarted blog. A lot of views and reads (that also on 31st night.. I must say I have very cool & eager readers.. I hope everyone had a great new year's night..  

Math is everywhere.. A lot of my readers might remember my old orchestration on Ant Colony Optimization, Swarm Intelligence & Quantum-behaved Particle Swarm Optimization algorithm (And a fun hint on How Ant Colony Optimization is used in CyberSecurity with reference to first Avengers movie dialogue - Nick Fury to Loci when he is in the prison- "the touch of the ant and the click" ...

And And And... funnier hint that Marvel might release a movie on Ant-Man etc..

A lot of advances have happened in Swarm Intelligence since then.... 

1. Quantum Particle swarm optimization (PSO), 

2. Firefly algorithm (FA), 

3. Chicken swarm optimization (CSO), 

4. Grey wolf optimizer (GWO), 

5. Sine cosine algorithm (SCA), 

6. Marine-predators algorithm (MPA)

7. Archimedes optimization algorithm (AOA)

8. And Surprise Surprise - Duck Swarm Algorithm 

Please check out 2021 paper - Duck swarm algorithm: a novel swarm intelligence algorithm by Zhang , Wen, Yang - South China University of Technology, Guangzhou 

A couple of simple images from the paper - 

And the Pathway Analysis based on duck swarm - 

By the way, A very very interesting thing happened a few months back. Google came up with their own AI tool on pathway analysis - https://blog.google/technology/ai/introducing-pathways-next-generation-ai-architecture/  

The question is would Google use more & more swarm intelligence in their pathway analysis?