| viXra.org > Artificial Intelligence > 2411 |
 |
 |
| viXra.org > Artificial Intelligence > 2411 |
|
|
[13] viXra:2411.0162 [pdf] submitted on 2024-11-26 20:46:24
Authors: Siddharth Anand Phatak
Comments: 7 Pages.
This report presents the development and evaluation of a machinelearning model for identifying vulnerable C code. Using an AI-generateddataset of both vulnerable and non-vulnerable C code snippets, we explorevarious methodologies including Bag of Words (BOW), Logistic Regres-sion, word embeddings, and Recurrent Neural Networks (RNNs) to buildan effective classification model.
Category: Artificial Intelligence
[12] viXra:2411.0154 [pdf] submitted on 2024-11-25 21:54:10
Authors: Fei Ding
Comments: 6 Pages.
The recent proliferation of so-called open-source large language models (such as LLaMA, Falcon, Mistral) has introduced a broader range of alternatives for AI practitioners and researchers. However, the majority of these models cannot be considered truly open-source, as they often provide only partial artifacts, such as final model weights or inference code. Furthermore, technical documentation accompanying these models tends to focus on high-level architectural decisions and superficial metrics, leaving critical aspects of the training process, including dataset composition, distribution, model checkpoints, and intermediate results, largely undisclosed. This lack of transparency presents a significant barrier to progress in the field, restricting the potential for open, collaborative research. In the absence of access to original datasets, attempts to further train or fine-tune these models by third parties are susceptible to issues such as catastrophic forgetting.In response to this challenge, we propose a method that facilitates more effective supervised fine-tuning of these closed-source models, without requiring access to the original data, while mitigating the risk of catastrophic forgetting.
Category: Artificial Intelligence
[11] viXra:2411.0124 [pdf] submitted on 2024-11-19 11:56:05
Authors: Tofara Moyo
Comments: 3 Pages.
We present a novel scientific document discoverysystem inspired by molecular chemistry and AI-driven drug discovery. Our approach treats document tokens as atomic units, which are combined to form "molecular" representations ofmathematical documents. We employ a probabilistic framework to maximize the likelihood of forming coherent mathematicaldocuments while minimizing the probability of random token combinations and non-STEM document tokens. To achieve this, we develop a token embedding scheme that maps property vectors to a musical keyboard, effectively representing each token as a musical chord. We further differentiate between STEM and non-STEM documents by introducing a harmonic constraint on adjacent nodes in document graphs. Specifically, STEM documents are characterized by polyphonic harmonization of adjacent node vectors, whereas non-STEM documents exhibit dissonant relationships. Our system integrates a graph neural network/transformer decoder architecture, trained end-to-end to generate STEM documents from input graphs. This innovative approach has the potential to revolutionize scientific document discovery and retrieval.
Category: Artificial Intelligence
[10] viXra:2411.0116 [pdf] submitted on 2024-11-17 16:02:08
Authors: Tofara Moyo
Comments: 5 Pages.
We present a novel method for learning hierarchical abstractions that prioritize competing objectives, leading to improved global expected rewards. Our approach employs a secondary rewarding agent with multiple scalar outputs, each associated with a distinct level of abstraction. The traditional agent then learns to maximize these outputs in a hierarchical manner, conditioning each level on the maximization of the preceding level. We derive an equation that orders these scalar values and the global reward by priority, inducing a hierarchy of needs that informs goal formation. Experimental results on the Pendulum v1 environment demonstrate superior performance compared to a baseline implementation.We achieved state of the art results.
Category: Artificial Intelligence
[9] viXra:2411.0102 [pdf] submitted on 2024-11-13 22:17:11
Authors: Xiaoyi Li
Comments: 10 Pages.
Generative AI models are increasingly used across various modalities, including text, images, audio, and video. Estimating the computational cost of generating con- tent is crucial for optimizing performance and resource allocation. This paper intro- duces the Cost-Per-Byte Principle: C = T × I, a universal law that relates the cost of content generation to per-byte generation time and per-second inference cost. We derive the per-byte generation time analytically based on the model’s computational requirements (FLOPs) and the hardware’s performance (FLOPs per second). By estab- lishing mappings between bytes and different content units (characters, pixels, samples, frames), we provide a modality-agnostic framework for cost estimation. We present a rigorous proof of the principle’s validity and apply it to estimate the costs of current popular models, using publicly available evidence to verify the accuracy and usefulness of this principle.
Category: Artificial Intelligence
[8] viXra:2411.0090 [pdf] submitted on 2024-11-12 03:39:38
Authors: Mezbah Uddin Rafi
Comments: 17 Pages.
Emotional intelligence (EI) is crucial for interpersonal interactions, mental health, and success across various life domains. Traditionally enhanced through coaching, workshops, and self-guided methods, EI development can now leverage artificial intelligence (AI) as a virtual emotional coach. With advancements in machine learning (ML), natural language processing (NLP), and sentiment analysis, AI can offer real-time emotional assessment and personalized feedback, providing an innovative approach to EI training.
Category: Artificial Intelligence
[7] viXra:2411.0083 [pdf] replaced on 2025-04-02 21:21:33
Authors: Ait-Taleb Nabil
Comments: 7 Pages.
In this paper, we will expose for the Gaussian multiple a theorem relating the predictability to correlations. This theorem is based on another equality which will be also proven. For the correlations to be predictability, the proof will show that the variance-covariance matrix must be located onto the boundary of the positive semi-definite matrix cone with only one zero eigenvalue.
Category: Artificial Intelligence
[6] viXra:2411.0057 [pdf] submitted on 2024-11-07 02:17:57
Authors: Gopal Krishna
Comments: 5 Pages.
This paper establishes the fundamental nature of general intelligence and proves the logical impossibility of Artificial General Intelligence (AGI). We introduce the novel framework of Abstract Sentient Intuition (ASI) and Combinatorial Sentient Intuition (CSI), demonstrating that while CSI involves combining existing abstract concepts, ASI creates fundamentally new abstract concepts. Building upon the established foundation of abstract language, we prove that artificial systems can only implement CSI through programming, as all programming is fundamentally based on existing knowledge. Since general intelligence requires both ASI and CSI, we establish that AGI is logically impossible. We systematically address all potential counterarguments, demonstrating the completeness of this proof. This result has profound implications for artificial intelligence research, cognitive science, and our understanding of consciousness.
Category: Artificial Intelligence
[5] viXra:2411.0031 [pdf] submitted on 2024-11-04 08:48:35
Authors: Eugene Rulko
Comments: 18 Pages.
The main hurdle for terrain relative navigation systems is the incongruity of visual features between a patch of a satellite reference map and a view from an onboard UAV camera. Images are taken during different time of year, under different weather, vegetation and lighting conditions, with different angles of observation. This work proposes the usage of deep feature template matching, where features are extracted during unsupervised training using a triplet loss. It provides semantic understanding, agnostic to terrain transformations. In order to overcome struggling to navigate over featureless terrains, the work proposes additional usage of visual odometry with the procedure of sticking to the map after encountering enough features, with the procedure of hypothesizing over possible locations. Passing a fragment of the reference map through the trained feature extractor, applying an entropy filter and then a pathfinding algorithm allows planning a flying path over areas rich of features relevant for navigation.
Category: Artificial Intelligence
[4] viXra:2411.0029 [pdf] submitted on 2024-11-04 10:40:36
Authors: Mirtill Boglárka Naghi, Bence Tureczki, Katalin Szenes
Comments: 23 Pages.
This paper introduces a novel approach to fortify data security through the seamless integration of fuzzy clustering techniques within blockchain technology. Fuzzy clustering, known for its ability to handle uncertainties and complexities in data, synergizes with blockchain’s decentralized and immutable ledger to establish a robust framework for secure data storage, analysis and retrieval. The proposed fusion not only enhances confidentiality, integrity and effectivity but also offers adaptability to the evolving dynamics of modern data landscapes. In this paper we propose a theoretical model that implements the integration of fuzzy c-means clustering on the blockchain using a cryptographically verifiable distributed computing system. By leveraging the decentralized nature of blockchain, the proposed framework ensures that data analysis processes are verifiable and tamper-resistant. Furthermore, the integration of fuzzy clustering within the blockchain not only bolsters security but also introduces a layer of transparency in the confidential data handling process.
Category: Artificial Intelligence
[3] viXra:2411.0021 [pdf] submitted on 2024-11-03 02:18:46
Authors: Oleg Kupervasser, Domoshnitsky Alexander
Comments: 45 Pages.
In the presentation described algorithms for airborne ground robot's control and navigation developed in Ariel University during Kamin project
Category: Artificial Intelligence
[2] viXra:2411.0020 [pdf] submitted on 2024-11-03 02:23:41
Authors: Oleg Kupervasser, Domoshnitsky Alexander
Comments: 72 Pages.
In the presentation described algorithms for Vision-based UAV (Unmanned aerial vehicle) control and navigation developed in Ariel University during Nofar project.
Category: Artificial Intelligence
[1] viXra:2411.0004 [pdf] submitted on 2024-11-01 20:37:47
Authors: Vitaly E. Pilkin
Comments: 12 Pages. (Correction made by viXra Admin to conform with the requirements of viXra.org)
This paper provides answers to current questions of experts working with artificial intelligence (AI), and offers recommendations on how to control AI development and prevent AI from getting out of human control.
Category: Artificial Intelligence
| Third-Party Links: |
|