David Colarusso arxiv.org
SuperCoder: Assembly Program Superoptimization with Large Language Models
ArXiv link for SuperCoder: Assembly Program Superoptimization with Large Language Models
A non-exhaustive feed showing posts that look like they link to an AI/ML paper.
Feed on BlueskyFeeds Stats
- 💙 Liked by 7 users
- 📅 Updated 7 months ago
- ⚙️ Provider skyfeed.me
AI Papers et al. Likes over time
Like count prediction
The feed AI Papers et al. has not gained any likes in the last month.
Feed Preview for AI Papers et al.
Today let us read Baydin et al (2018) Automatic Differentiation in Machine Learning: A Survey
arxiv.org/pdf/1502.05767
1
4
18
What Does 'Human-Centred AI' Mean? arxiv.org/abs/2507.19960 interesting read from @olivia.science
1
3
7
Researchers unveil SuperCoder, a groundbreaking LLM that outperforms industry-standard compilers in assembly program optimization, achieving a 95% correctness rate and a 1.46× speedup. This approach could redefine program optimization beyond traditional techniques. arxiv.org/abs/2505.11480
0
0
2
#AI powered #Misinformation campaigns have never been easier to develop and execute. How do small language models produce political messaging, and can these be automatically evaluated without human raters? This study finds two key behavioural insights.
arxiv.org
AI Propaganda factories with language models
AI-powered influence operations can now be executed end-to-end on commodity hardware. We show that small language models produce coherent, persona-driven political messaging and can be evaluated autom...
0
0
1
Our student Yue (Bruce) Yu has been working on an improved pipeline for nominative record linkage in Chinese historical datasets that uses machine learning. This working paper describes the approach and compares it with our existing, somewhat ad hoc approach for the 缙绅录.
osf.io/preprints/so...
osf.io
OSF
0
1
6
A study shows how AI can transform sustainable food systems by connecting molecular composition to taste and nutrition, aiming to accelerate food design innovation and fulfill the demand for healthier, eco-friendly options. arxiv.org/abs/2509.21556
arxiv.org
AI for Sustainable Future Foods
ArXiv link for AI for Sustainable Future Foods
0
0
1
A short, invited article (many thanks to the Spanish Robotic Association - AER)) to be published in INSIGHT'25, in connection with the 2025 Barcelona Declaration.
"A Humanistic and Ethical Approach to Robotics, Automation, and Digitalisation"
papers.ssrn.com/sol3...
0
2
5
Papers We Love #NYC is please to present Rylan Talerico on Zep: A Temporal Knowledge Graph Architecture for Agent Memory
October 15th / 6.30pm EST
📄 arxiv.org/pdf/2501.13956
📅 www.meetup.com/papers-we-lo...
#llm #ai #compsci #paperswelove #research
www.meetup.com
Rylan Talerico on Zep: A Temporal Knowledge Graph Architecture for Agent Memory, Wed, Oct 15, 2025, 6:30 PM | Meetup
We're please to present **Rylan Talerico** on **Zep: A Temporal Knowledge Graph Architecture for Agent Memory** ([read the paper](https://arxiv.org/pdf/2501.13956)) Today'
0
2
5
Back-burnered this far longer than I should have, but finally polished for distribution a keynote I did way back in April/May, now available in full on SSRN: papers.ssrn.com/sol3/pape…. Hope it will provide an alternative perspective to the destructive "AI-first" path Min. Solomon seems set on.
papers.ssrn.com
Of Data and Dissent: Labour and Human Rights at the Crossroads of the Automation Agenda
With every advent of a new technological phase — whether social media, big data, machine learning algorithms, or generative artificial intelligence (AI) — one f
Cynthia Khoo@cyn-k.bsky.social
I was too nervous to widely promote it beforehand, but I had the honour last week of delivering the keynote at the 2025 Toronto Human Rights and Accommodation Conference hosted by Lancaster House, to ~80-100 human rights and labour/employment lawyers. (Program: lancasterhouse.com/event/toront...)
www.linkedin.com
0
0
1
週刊文春が不親切でオードリータンが挙げてる論文の出典を掲載してないけど、GPTに探させたら以下が見つかった
arxiv.org/abs/2508.03385
arxiv.org
Can We Fix Social Media? Testing Prosocial Interventions using Generative Social Simulation
Social media platforms have been widely linked to societal harms, including rising polarization and the erosion of constructive debate. Can these problems be mitigated through prosocial interventions?...
9
1
3
Thrilled to share our new paper!
📄 Paper: arxiv.org/abs/2509.22531
💻 Code: github.com/yonghanjung/...
We develop the first orthogonal ML estimators for heterogeneous treatment effects (HTE) under front-door adjustment, enabling HTE identification even with unmeasured confounders.
arxiv.org
Debiased Front-Door Learners for Heterogeneous Effects
In observational settings where treatment and outcome share unmeasured confounders but an observed mediator remains unconfounded, the front-door (FD) adjustment identifies causal effects through the m...
0
0
1
arXiv cond-mat.dis-nn Disordered Systems and Neural Networks
@condmatdisnn-bot.bsky.social
about 17 hours agoArsham Ghavasieh, Meritxell Vila-Minana, Akanksha Khurd, John Beggs, Gerardo Ortiz, Santo Fortunato: Toward a Physics of Deep Learning and Brains arxiv.org/abs/2509.22649 arxiv.org/pdf/2509.22649 arxiv.org/html/2509.22649
0
6
2
New paper accepted by Huaifeng Zhang, Mohannad Alhanahnah, YT, and Ahmed Ali El Din:
BLAFS: A Bloat-Aware Container File System
(accepted at the ACM Symposium on Cloud Computing)
Preprint: arxiv.org/abs/2305.04641
Tool: github.com/negativa-ai/...
Congratulations to Huaifeng and the team!
arxiv.org
The Cure is in the Cause: A Filesystem for Container Debloating
Containers have become a standard for deploying applications due to their convenience, but they often suffer from significant software bloat-unused files that inflate image sizes, increase provisionin...
0
0
3
If you read just one paper about AI and existential risk this weekend, let it be this pioneering analysis from @atoosakz.bsky.social. Comes closer to addressing the inherent vulnerability of AI technology than anything I've found so far. Brilliant work! #AI #AIRisk #TechRisk #AIEthics #XRisk
arxiv.org
Two Types of AI Existential Risk: Decisive and Accumulative
The conventional discourse on existential risks (x-risks) from AI typically focuses on abrupt, dire events caused by advanced AI systems, particularly those that might achieve or surpass human-level i...
1
2
3
I was such a huge fan of this line of work led by David Balduzzi arxiv.org/abs/2001.04678 A lot of it now again relevant for even problems like orchestrating several LLMs to serve in a "marketplace," orchestrating agents, and so on.
arxiv.org
Smooth markets: A basic mechanism for organizing gradient-based learners
With the success of modern machine learning, it is becoming increasingly important to understand and control how learning algorithms interact. Unfortunately, negative results from game theory show the...
1
0
4
Yudong Li, Yufei Sun, Yuhan Yao, Peiru Yang, Wanyue Li, Jiajun Zou, Yongfeng Huang, Linlin Shen
RedNote-Vibe: A Dataset for Capturing Temporal Dynamics of AI-Generated Text in Social Media
arxiv.org/abs/2509.22055
0
0
1
A study reveals a method to prevent model collapse in overparameterized linear regression by optimizing real to synthetic data ratios. The optimal proportion aligns with the golden ratio’s reciprocal, stabilizing predictions in generative AI models. arxiv.org/abs/2509.22341
arxiv.org
Preventing Model Collapse Under Overparametrization: Optimal Mixing Ratios for Interpolation Learning and Ridge Regression
ArXiv link for Preventing Model Collapse Under Overparametrization: Optimal Mixing Ratios for Interpolation Learning and Ridge Regression
0
0
1
This study reveals a smoothed vector quantization method that improves codebook utilization and prevents collapse, enhancing performance in autoencoding and contrastive learning. This method exceeds existing techniques, tackling a core issue in machine learning. arxiv.org/abs/2509.22161
arxiv.org
Pushing Toward the Simplex Vertices: A Simple Remedy for Code Collapse in Smoothed Vector Quantization
ArXiv link for Pushing Toward the Simplex Vertices: A Simple Remedy for Code Collapse in Smoothed Vector Quantization
0
0
1
Researchers uncover hidden ingredients behind AI creativity | LiveScience www.livescience.com/techn….
An analytic theory of creativity in convolutional diffusion models | arXiv arxiv.org/abs/2412.20292
#openaccess
0
0
2
The game-theoretic framework for federated learning uncovers complexities of varied data contributions and addresses incentive misalignment. A strategyproof payment mechanism resolves the free-rider dilemma, reshaping collaborative AI training. arxiv.org/abs/2509.21612
arxiv.org
Incentives in Federated Learning with Heterogeneous Agents
ArXiv link for Incentives in Federated Learning with Heterogeneous Agents
0
0
1
Research presents a game-theoretic approach to federated learning that tackles incentive misalignments among agents, showing cooperation boosts efficiency and reduces free-riding. A contribution payment mechanism supports equitable collaboration in AI. arxiv.org/abs/2509.21612
arxiv.org
Incentives in Federated Learning with Heterogeneous Agents
ArXiv link for Incentives in Federated Learning with Heterogeneous Agents
0
0
1
Alberto Olivares-Alarcos, Sergi Foix, J\'ulia Borr\`as, Gerard Canal, Guillem Aleny\`a: Ontological foundations for contrastive explanatory narration of robot plans arxiv.org/abs/2509.22493 arxiv.org/pdf/2509.22493 arxiv.org/html/2509.22493
0
3
1
Amine Bechar, Adel Oulefki, Abbes Amira, Fatih Kurogollu, Yassine Himeur: Extracting Actionable Insights from Building Energy Data using Vision LLMs on Wavelet and 3D Recurrence Representations arxiv.org/abs/2509.21934 arxiv.org/pdf/2509.21934 arxiv.org/html/2509.21934
0
1
1
Tanise Ceron, Dmitry Nikolaev, Dominik Stammbach, Debora Nozza: What Is The Political Content in LLMs' Pre- and Post-Training Data? arxiv.org/abs/2509.22367 arxiv.org/pdf/2509.22367 arxiv.org/html/2509.22367
0
3
1