使用区块链原理改进人工智能 研究和安全.docx
版权申诉
38 浏览量
2024-04-18
00:42:21
上传
评论
收藏 552KB DOCX 举报
1
D
D
Using blockchain principles for improving AI
research and security
Kresimir Kalafatic,
E-mail:kresimir.kalafatic@gmail.com
Abstract—Distributed ledger (blockchain) is a distributed database whose every written record is signed by private key, every record
insertion and change can be traced back to specific public/private key pair and whose atomic operation is irrecoverably committed after
the databases mathematically prove their validity and distributed consistency. The mathematical principle of blockchains can be used in
improving AI research and security.
Index Terms—blockchain, AI, security, linking research data and code
令
1
INTRODUCTION
ISTRIBUTED LEDGER (BLOCKCHAIN) is a distributed
database whose every written record is signed by
private key, every record insertion and change can be traced
back to specific public/private key pair and whose atomic
operation is irrecoverably committed after the databases
mathematically prove their validity and distributed con-
sistency. The blockchain is based on statistical and crypto-
graphic principles for improving security of data manage-
ment enabling authorized and audited change and process-
ing of data from its origin to the end of data usage.
Cybersecurity is an important element in every sector
of human interaction and lately regulators are introducing
new frameworks for increasing baseline security of all insti-
tutions under their supervision. Even though formal proce-
dures are used in software and hardware component design,
most of the designs have security flaws which have their
origin in errors of human personnel or computer models. To
reduce the security flaws the design origin has to be proven
by mathematics and physics.
Knowing the origin of data enables ranking the data
quality of different data sources in machine learning. Sen-
sors can experience degradation during their lifetime and
blockchains enable knowing which contaminated training
data has to be discarded or repaired. Using multiple parallel
sensors performing the same task and using appropriate
consensus protocols can reduce the influence of contami-
nated data sources on the final AI decision.
Having a decentralized database improves data avail-
ability and infrastructure resiliency while enabling creation
of different multiple AI designs using the same replicated
datasets with different splits for training, validation, test-
ing datasets. Decentralized computer models enable using
distributed computer resources and splitting the work into
smaller parts enabling resource polling architecture.
2
SHORT OVERVIEW OF SOME BLOCKCHAIN PRIN-
CIPLES
ISTRIBUTED LEDGER (BLOCKCHAIN) implementations
are in different stages of development and some of the
most popular implementations are bitcoin and ethereum.
There are many papers describing the principles of bitcoin
and ethereum, so this paper will concentrate on just few
important principles.
2.1
Public/Private keys and wallet address
Public and private keys are based on cryptographic prin-
ciples. Bitcoin and ethereum use the Elliptic Curve Digital
Signature Algorithm. Public/Private-key cryptography en-
ables digitally signing data with a private key and anyone
who knows the public key of an entity can verify that the
signature is valid.
The asset (data or some other) is assigned to the owner
based on a wallet address. The wallet address is derived
using hash algorithm on public key. The wallet address adds
additional layer of security because the public key is hidden
until the transaction is initiated from the wallet. Using wal-
let addresses enables data anonymization which is required
by the EU GDPR (General Data Protection Regulation).
2.2
Decentralized database consistency algorithms
Atomic operation in blockchain is irrecoverably commit-
ted after the databases mathematically prove their validity
and distributed consistency. There exist different types of
consensus algorithms which form several groups. ”Proof
of work” and ”Proof of stake” consistency algorithms are
based on mathematics and operate on economic principles
and monetary metrics. Generation of new blocks (tokens)
is used for securing database consistency and history while
enabling covering operational cost and investment in the
infrastructure by selling mined blocks to the network mem-
bers.
2.2.1
Proof of work
Bitcoin and ethereum are currently operating using ”Proof
of work (PoW)” consensus. Bitcoin ”Proof of work” is based
on autocorrelation function which uses hash of previously
generated block, blocks from latest transactions, address
of the miner and other data. The autocorrelation function
资源评论
百态老人
- 粉丝: 1311
- 资源: 2万+
下载权益
C知道特权
VIP文章
课程特权
VIP享7折，此内容立减5.97元 开通VIP
最新资源
- 基于matlab实现夜间车牌识别程序(1).rar
- 基于matlab实现图像处理，本程序使用背景差分法对来往车辆进行检测和跟踪.rar
- 基于matlab实现视频监控中车型识别代码，自己写的，希望和大家多多交流.rar
- sdk.config
- 基于matlab实现配电网三相潮流计算方法，对几种常用的配电网潮流计算方法进行了对比分析.rar
- 基于matlab实现配电网潮流 经典33节点 前推回代法潮流计算 回代电流 前推电压 带注释.rar
- 基于matlab实现模拟退火遗传算法的车辆调度问题研究,用MATLAB语言加以实现.rar
- 基于matlab实现蒙特卡洛的的移动传感器节点定位算法仿真代码.rar
- 华中数控系统818用户说明书
- 基于matlab实现卡尔曼滤波器完成多传感器数据融合 对多个机器人的不同传感器数据进行融合估计足球精确位置.rar
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
