A blog on social software, decentralized identity, and machine learning by Cristian Lepore.

Toward Self-Sovereign Identity

May 2021 - 4000 Words

Blockchain started as the backbone for cryptocurrencies and emerged as one ofthe most relevant technologies of the last decade. Today blockchain has numerous fields of application, one of these is in the identity management landscape. Identity management is the process to manage users’ identities and provide accessto technology resources. This process is gaining importance due to the growing number of digital identities that we deal with. As an increasing number of peoplehave been spending time online and as more activities have been drifting toward the digital world due to the recent pandemic, the notion of identity slided from the physical world to the digital world as well. The new idea of identity is inter-esting and complex at the same time and it represents the digital fingerprint ofa person in the digital space. Who you are with respect to how other see you, isa relevant question that we want to address when giving a definition of digital identity.

Read more →

A Survey on Blockchain Consensus with a Performance Comparison of PoW, PoS and Pure PoS

January 2021 - 14000 Words


Blockchain has emerged as one of the most interesting technologies of the last decade. Today, the industries' concern is about their ability to handle a high volume of data transactions per second while preserving both decentralization and security. Many blockchains differentiate theirselves based on the consensus method that they employ. This article presents a review of the main consensus procedures and provides a framework to compare the performances of PoW, PoS, and the Pure PoS, based on throughput and scalability.

Read more →

Using Neural Network to Intrusion Detection System

September 2020 - 4500 Words


Intrusion Detection System (IDS) predominantly works for detecting malicious attacks on a computer network. In this project, we propose some IDS models based on Artificial Neural Network and clustering techniques. ANNs can only be built if there is the availability of an effective dataset. A dataset with a sizable amount of quality data that mimics the real world can help to train and test an intrusion detection system. We will introduce the NSL-KDD dataset. It has been used to study the effectiveness of the neural network-based classification algorithm to detect anomalies in the network traffic patterns.

Read more →

Performance Analysis through Adaptive Boosting

March 2020 - 5500 Words


This paper shows the results on risk analysis and overfitting, obtained thanks to parametric and non-parametric learning algorithms. The proposed classifiers are decision trees, decision stumps, and perceptrons. In order to evaluateperformance improvement, all predictors were boosted with an ensemble learning algorithm. The Breast Cancer database was chosen as the dataset for the tests, freely downloadable from the UCI repositories.

Read more →

Algorand Consensus in a Nutshell

November 2019 - 3500 Words


Algorand is a novel blockchain technology introduced by 2015 by the Turing award Silvio Micali which claims to solve the Trilemma problem. The term algorand is obtained by mixing together two distinct words: algorithmic and randomness. These two words suggest that Algorand is based on both a mathematical approach and a random process. The algorithmical part and the randomness are both involved in the process of generating a pseudo-random1 number as proposed by Micali, Rabin, and Vadhan. Later, Micali noticed that this work could be useful to cope with the blockchain’s idea to randomly (and privately) select a leader and imported this pseudo-random generator to be part of the Algorand blockchain.

Read more →

Presenting the Byzantine Generals' Problem

July 2019 - 600 Words


The Byzantine Generals’ Problem is a practical solution to the Byzantine faults’ problem given by Lamport in 1982. The best way to explain the analogy is through a historic tale. The depicted scenario consists of an army that has besieged the city. The army has n divisions with n > 0, and each division has a lieutenant. In particular, there is a commanding general plus n lieutenants that are close to the enemy target. The commanding general must take a decision whether to attack or retreat and send the order to the other lieutenants. Be aware that the army has a good chance to win if and only if a majority of 2/3 of the lieutenants agree to the same overall decision (attack or retreat).

Read more →

Profle picture
  • Cryptography
  • Digital Identity
  • Blockchain [Article]

GitHub - LinkedIn
© Cristian Lepore