Keynotes and Talks Summary: NORA AI Annual Conference 2023, Tromsø, Norway
Published on June 12, 2023 (Funding by HiØ)
I am currently a Master's student (2022-24) at Østfold University College and a summer Intern (2023) at OsloMet University, Norway. My research focus belongs to ALife, Evolution, Complexity and Artificial General Intelligence. I enjoy working with Prof Stefano Nichele (also supervisor) where we together study complexity and emergent behaviour in discrete Cellular Automata, continuous CA (Lenia) and Neural CA. Some of my previous work and projects also determine my interest in Deep Learning for Computer Vision and Explainable AI. I finished my Junior Research Fellowship (and did not continue PhD) at Indian Institute of Technology (IIT) Jodhpur (2020-2022) where I studied Deep Learning, Classical ML, and Dependable AI taught by Dr. Mayank Vatsa and Dr. Richa Singh. Conversely, I was also part of VANETs lab supervised by Prof Debasis Das, where my focus was to provide Machine Learning edge for the team. Moreover, I was part of QIC group at the IIT as well. I finished my early Computer Science Engineering (Bachelor's 2015-2019) degree from UPES, Dehradun at the foothills of Himalayas. Most of my guidance during bachelor's for which I am thankful, always to Dr Ravi Tomar. I belong to the central part of India, from a town called Banswara which lies in south Rajasthan.
Jain, S., Shrestha, A., Nichele, S. (2023). Capturing Emerging Complexity in Lenia. arXiv preprint arXiv:2305.09378.
Jain, S. (2023). Adversarial Attack On Yolov5 For Traffic And Road Sign Detection. arXiv preprint arXiv:2306.06071.
Jain, S. (2023). DeepSeaNet: Improving Underwater Object Detection using EfficientDet. arXiv preprint arXiv:2306.06075
Jain, S. (2023). CQural: A Novel CNN based Hybrid Architecture for Quantum Continual Machine Learning. arXiv preprint arXiv:2305.09738.
Jain, S. (2023). ADDSL: Hand Gesture Detection and Sign Language Recognition on Annotated Danish Sign Language. arXiv preprint arXiv:2305.09736.
Jain, S. (2023). RAMario: Experimental Approach to Reptile Algorithm--Reinforcement Learning for Mario. arXiv preprint arXiv:2305.09655.
Frequency Histogram Coarse Graining and MNCA - Sanyam Jain, Stefano Nichele
Studying Complexity and Emerging Behaviour in ECA, Discrete CA and 2D MNCA. Cellular automata and other discrete dynamical systems have long been studied as models of emergent complexity. We argue that being able to capture emergent complexity in AI systems may pave the way to open-ended evolution, a plausible path to reach artificial general intelligence.
MNCA Portal
Paper Released!
EvoLenia Portal - Sanyam Jain, Aarati Shrestha, and Stefano Nichele
Capturing emerging complexity in Lenia using EC fitness based on VoT, AE and AEVoT. This work investigates the emergent complexity in Lenia, an artificial life platform that simulates ecosystems of digital creatures. Le- nia’s ecosystem consists of a continuous cellular automaton where simple artificial organisms can move, grow, and reproduce. Overall, this project aims at investigating the potential of Lenia as ecosystem for emergent complexity in open-ended artificial intelligence systems.
Evo-Lenia Portal
Paper
DeepSeaNet Portal: Using EfficientDet object detection for Denmark's Brackish-Dataset
The results indicate that modified EfficientDet achieved higher accuracy with five-fold cross validation than the other models with 88.54% IoU of feature maps.
DeepSeaNet Portal
Paper
Adversarial Attack On Yolov5 For Traffic And Road Sign Detection
The paper investigates the impact of different types of attacks, including the Limited memory Broyden Fletcher Goldfarb Shanno (L-BFGS), the Fast Gradient Sign Method (FGSM) attack, the Carlini and Wagner (C&W) attack, the Basic Iterative Method (BIM) attack, the Projected Gradient Descent (PGD) attack, One Pixel Attack, and the Universal Adversarial Perturbations attack on the accuracy of YOLOv5 in detecting traffic and road signs.
GitHub Repository
Paper
Published on June 12, 2023 (Funding by HiØ)
Published on June 10, 2023 (Funding by HiØ)
Published on November 16, 2022 (Funding by HiØ)
Published on September 11, 2023 (Funding by HiØ)
Published on May 7, 2024 (Funding by HiØ)
Here, you can find information about my professional experience.
Degree: Masters in Science
Major: Applied Computer Science
Year: 2022-2024
Currently maintaining 4.4/5 - Stduying Machine Learning specialisation with CPS. Projects for the courses adv. ML and Applied CS project includes "DeepSeaNet" and "Complexity in Lenia" which can be found in publications section as part of preprint.
Click for more!Degree: PhD (Dropped)
Major: Quantum Information and Computation with focused coursework on Deep Learning
Year: 2020 - 2022
Grade: 7.8/10, Activities and societies: Part of VANETS LABActivities and societies: Part of VANETS LAB, Courses: Machine Learning, Dr G. Harit AI 1, Dr D. Mishra Graph Theory and Applications, Dr A. Mishra Deep Learning, Dr M. Vatsa Dependable AI, Dr R. Singh (A- grade) Financial Engg., Dr V. Vijay (A- grade) Cryptography, Dr S. Sanadhya ML for Economics, Dr D. Brahma Adv Machine Learning, Dr M. Vatsa (A grade) Cryptography, Course Project Theoretical analysis of QKD protocol BB84 and experimenting related algorithms on qiskit, (Endsem) QKD protocols and Quantum attacks on blockchain. (Attacks on Bitcoin PoW and Attacks on Digital Signatures) ML2, Endsem Project - Analyzing and Improving the Image Quality of StyleGAN (StyleGAN2) CVPR 2020, HELEN EYE DATASET and Clothing Dataset (Fashion Dataset) ML2, Group Project - Random Multimodel Deep Learning for Classification (arXiv:1805.01890) 20NewsGroup and IMDB dataset & CIFAR-10, MNIST Dataset. Dependable AI - Study of Nature of Different Adversarial Attacks and Impact of Different Data Augmentation Techniques on Robustness.
Click for more!Degree: BTech (Bachelor's in Technology)
Major: Computer Science Engineering
Year: 2015 - 2019
Grade: 77.50, Best Project award for Compariso - whatsapp bot to compare e-commerce prices. Top 10 in R.I.S.E. UPES under supervision of Dr. Ravi Tomar sir with project title: KidExa (Team KeyringCorp) Won Best project Award in Project Parliament (2018) college Minor Project 2 for Facial Recognition significance in cybersecurity. (Team KeyringCorp) Won Third Prize for Real-time Traffic Analysis – IBM ICE on campus Hackathon. (Team KeyringCorp) Won Second prize for 4FactorAuthentication in North India Cyber Security Hackathon at WIC Dehradun. (Team KeyringCorp) Developed Python Chatbot using wit NLP engine. (Wit.ai) Developed Automated Key logger using C. (Unix/Py) Studied Digital Logic, Data Structures and Algorithms, Java, Operating Systems, DBMS, Computer Networks, Theory of Computation, Discrete Maths, Information Retrieval and Search engines, Artificial Intelligence, Mobile Applications.
Click for more!You can reach me via email at sanyamjaincs@gmail.com. Connect with me on social media through the following profiles:
Ready to serve customers! Our poster presented by @s4nyam @hiofnorge #NordicAIMeet2023 https://t.co/J0tzVaT2Tg pic.twitter.com/EL44fTbANF
— Stefano Nichele (@stenichele) October 2, 2023
Shared some ideas, discussion, and conversation about ALife, Complexity and OpenEndedness over Conference dinner with Shin (linkedin below)😊#NORDICAIMEET @NORAdotAI @AiCentreDK https://t.co/BCLNjWpd9c pic.twitter.com/RKe4hNLQIf
— Sanyam Jain (@s4nyam) October 2, 2023
.@hiofnorge is well represented at @NORAdotAI #NORDICAIMEET2023 with @s4nyam https://t.co/Td6eTg5fCH
— Stefano Nichele (@stenichele) October 2, 2023
Hello Venice😍 #WIVACE2023 https://t.co/YxbmGQMtXz pic.twitter.com/rSPhQZqo95
— Sanyam Jain (@s4nyam) September 5, 2023
A nice work from @Michielstock on “How to make everything: open-endedness in synthetic biology” at #WIVACE2023 @CaFoscari 👏 pic.twitter.com/bLmWjTUWvN
— Sanyam Jain (@s4nyam) September 8, 2023
Right now, @s4nyam presenting in #Wivace2034 his work on the AEVoT algorithm pic.twitter.com/ujUUfxadTe
— JJ Merelo (@jjmerelo) September 7, 2023
Ready for poster presentation at @NORAdotAI NORDIC AI MEET 2023, Copenhagen
— Sanyam Jain (@s4nyam) September 25, 2023
- with @stenichele, Aarati, @hiofnorge #NORA #Lenia #CellularAutomata pic.twitter.com/8eqjd0y2OK
A good sunny day from Halden, Norway 🌞 pic.twitter.com/Ds6BkgeCmH
— Sanyam Jain (@s4nyam) September 13, 2023
A step towards sexual reproduction between 🐠 and 🦎 with a single egg, in an extension of @sina_lana’s work. Also trying to increase training steps and updating perception model with @stenichele #NeuralCA pic.twitter.com/cmS1MGWm1f
— Sanyam Jain (@s4nyam) August 15, 2023
We are approaching this multi scale complexity (as degree of hierarchy) with coarse graining, which seems to approximate the "filtered" space-time diagrams
— Stefano Nichele (@stenichele) August 3, 2023
Work done with @s4nyam, a portal to visualise coarse grained computations for 1D and 2D CA coming soon :)
6/n, n=6
While studying EC these days, I came across some functions and their simulations:
— Sanyam Jain (@s4nyam) August 24, 2023
- Ackley Function
- Himmelblau's Function
- Rastrigin Function
- Rosenbrock Function
A thread 🧵#EvolutionaryComputation
Tried to solve those complex functions using #GeneticAlgorithm
Some quick theoretical insights and notes about open-endedness and complexification from the recent paper Biomaker CA:
— Sanyam Jain (@s4nyam) August 17, 2023
🧵#CellularAutomata #ArtificialGeneralIntelligence
https://t.co/7biiR7y2oq
— Sanyam Jain (@s4nyam) August 21, 2023
A simple EC based optimisation for f(x) = sin(x).cos(x) pic.twitter.com/LwaJdGcNe6
Returning from lab and this view 🌠 pic.twitter.com/UqAHEIDMDZ
— Sanyam Jain (@s4nyam) August 7, 2023
Happy to share that our work with @stenichele and Aarati has been accepted at WIVACE 2023. https://t.co/NDWgVK8h3q
— Sanyam Jain (@s4nyam) July 18, 2023
How to capture emerging complexity 🌀 in a 2D grid continuous cellular automata (Lenia) system using Genetic Algorithm 🧬 ?
— Sanyam Jain (@s4nyam) May 21, 2023
arXiv: https://t.co/403mHYQ9XQ 📝 with Aarati and supervision of @stenichele at @hiofnorge
A thread 🧵
Tried Hexagonal CA inspired from (Moore & Reile, 2004) @stenichele @cisne_hug
— Sanyam Jain (@s4nyam) June 26, 2023
The Simulation still uses known Replicator B/S[1,3,5,7]
Code: https://t.co/be6WiurMfD pic.twitter.com/xM9tEzQYCA
Mutating new creatures and complexity for #NAIM2023, Denmark with the Lenia project. #GeneticAlgorithm #Lenia #CellularAutomata with @stenichele and Aarati
— Sanyam Jain (@s4nyam) June 21, 2023
ref: https://t.co/403mHYQ9XQ pic.twitter.com/nAg8fDXpBu
Check out my latest article: Keynotes and Talks Summary: NORA AI Annual Conference 2023, Tromsø, Norway https://t.co/zIkowQUREp from @NORAdotAI
— Sanyam Jain (@s4nyam) June 12, 2023
with @stenichele and @hiofnorge
Preprints for the both the projects are live, have a look at them:
— Sanyam Jain (@s4nyam) June 12, 2023
For Applied CS Project: https://t.co/403mHYQ9XQ
For Adv ML Project: https://t.co/dvn7rXKqyu https://t.co/kAgUz1cEWu
Check out my latest article: NEURO AI Workshop, Tromsø, June 4, 2023 https://t.co/4HsZtq4cyM@stenichele @hiofnorge
— Sanyam Jain (@s4nyam) June 9, 2023
Ready with the poster for NORA 2023 Tromsø conference🚀 with @stenichele and Aarati.@NORAdotAI pic.twitter.com/hL4qHkx6ZP
— Sanyam Jain (@s4nyam) May 30, 2023
Lenia Kernel (K): Work as n’hood. Kernels 🎇 are blurred and handcrafted using gaussian smoothening function. Weighted importance to n’boring pixels and gradually reduced importance to distant pixels. Simply put, this is something that slides over space 🌐 to perform Convolution. pic.twitter.com/ew6tbc0gHQ
— Sanyam Jain (@s4nyam) May 21, 2023
We tested five-fold validation for these experiments to validate and understand it was not just a lucky case to get such fitness plot: pic.twitter.com/7Fv6QICMNF
— Sanyam Jain (@s4nyam) May 21, 2023
Finally ⭐, we initialised population with only known kernel and tested using AEVoT fitness for 500 generations and EC was able to find right set of perturbations using AEVoT fitness such that we were able to evolve a very interesting behaviour. Check arXiv preprint for details. pic.twitter.com/GZ36oJsn7Q
— Sanyam Jain (@s4nyam) May 21, 2023
To understand the importance of Gaussian Function in K, here is an intuitive animation that shows varying mu and sigma to develop a kernel. Note that sigma increases the spread of the peaks while mu displaces the peak. pic.twitter.com/yfgb2ab3Bm
— Sanyam Jain (@s4nyam) May 22, 2023
Had a chance to witness Norway constitution day and the parade, on the 17th may. 🇳🇴 pic.twitter.com/WopbAcsbTC
— Sanyam Jain (@s4nyam) May 17, 2023
An Evolutionary Computation EvoLenia run for 500 generations on default growth function and a handcrafted kernel shows a ring forming bacteria. Amazed by the robustness of Lenia behaviour, while mutating kernel variation based fitness.
— Sanyam Jain (@s4nyam) May 8, 2023
Part of project supervised by @stenichele pic.twitter.com/jLvl5E10s2
@s4nyam from @hiofnorge presenting SVM at the edge #ICAPAI2023 pic.twitter.com/jUrrnRrfWJ
— Stefano Nichele (@stenichele) May 2, 2023
Do we really need to kill our old plumber a million times for the princess?
— Sanyam Jain (@s4nyam) April 9, 2023
How about cheating, training and reaching only the Warp Zones in the game?#GymSuperMarioBros #AGI pic.twitter.com/UPwSfMkAt9