Artikkel avaldati algselt Novaatoris. Ehkki tehisaru teeb üha rohkem inimeste elu mõjutavaid otsuseid, ei suuda see neid alati inimesele arusaadavalt põhjendada. Tartu Ülikoolis kaitstud doktoritöö osutab, et inimesed ootavad masinalt realistlikke ja usaldusväärselt mõjuvaid seletusi. Tehisaru teeb järjest rohkem otsuseid, mis mõjutavad inimeste elu. Seda kasutatakse haiglates diagnooside toetamisel, pankades laenutaotluste hindamisel ning hariduses õppijate […]

Lecturer of Applied AI Armando Vieira shares his thoughts on Medium, an online publishing platform. The dominant discourse on AI alignment rests on a quiet but catastrophic assumption: that morality can be imposed on systems that do not experience anything. The debate unfolds as if values were transferable objects — encoded, optimized, constrained — rather than lived […]

The dominant paradigm in modern AI rests on a deceptively simple premise: intelligence emerges from minimizing prediction error across vast datasets. We train neural networks to be accurate, to match patterns, to reduce loss functions. Yet this approach, for all its empirical success, may be fundamentally misaligned with the nature of intelligence itself. Here is […]

You can find out more about Kristiina Rahkema’s PhD thesis in English here.  Nutitelefonidest on saanud meie elu lahutamatu osa. Iga asja jaoks on olemas äpp: sõnumite saatmiseks, internetipanga külastuseks, auto lukust avamiseks. Kahjuks on paljud neist rakendustest ebaturvalised. Turvavigu on avastatud sellistes rakendustes nagu Facebook, TikTok ja WhatsApp. Nende turvavigade raskusaste võib ulatuda andmelekkest […]

Introduction: Privacy-preserving computation is a crucial discipline that protects the privacy of confidential data while facilitating secure analysis. In graph problems, secure multiparty computation (SMC) presents a promising approach to performing these functionalities. This article delves into the notable advancements in privacy-preserving parallel algorithms recently introduced in a Ph.D. thesis. By integrating graph algorithms with […]

Kadi Tulver, arvutiteaduse instituudi kognitiivteaduse ja masinnägemise teadur Lugu sellest, kuidas Isaac Newton õunapuu all istudes murrangulise gravitatsiooniteooria peale tuli, on kõigile tuttav. Isegi kui mõned värvikad detailid on ilukirjanduslikult liialdatud, ilmestab see episood suurepäraselt ühte inimmõistusele unikaalset omadust – kuidas uudsed ideed mõnikord meile justkui tühja koha pealt pähe hüppavad. Püüdlustes luua tehisintellekti, mis […]

The cornerstone of democracy is the right for voters to participate in free and fair elections. However, securing elections is a non-trivial task due to the conflicting security requirements.  On the one hand, to provide the freedom to vote, it should not be possible to coerce voters into voting for a specific candidate. Therefore, the […]

Kaur Alasoo on Tartu Ülikooli arvutiteaduse instituudi bioinformaatika lektor. Tema artikkel ilmus Õhtulehes 22. mail 2022. Tegu on eesti noorte teaduste akadeemia algatatud artiklisarjaga “ElemENTAarne”, kus noorteadlased kirjutavad põnevatest teadustulemustest ja näitavad, kuidas teadus meie elu rikastab. Miks on nii, et teadlased suutsid vähem kui aastaga välja arendada mitu konkureerivat koroonaviiruse vastast vaktsiini, aga meil […]

Tartu Ülikooli arvutiteaduse instituudi keeletehnoloogia rühmas valmis bakalaureusetöö “Soome-ugri väikeste keelte neuromasintõlke edendamine” raames vabalt kasutatav mitmekeelne masintõlkesüsteem väiksematele soome-ugri keeltele nagu eesti, soome, võru, põhjasaami ja lõunasaami. Töö autor on Maali Tars ja teda juhendas Andre Tättar. Lõputöö põhjal avaldati teaduslik artikkel äsja toimunud NoDaLiDa keeletehnoloogia konverentsil pealkirja all “Extremely low-resource machine translation for […]

The original article (in Estonian) by Airika Harrik was published in Novaator. Estonia has used the ID card system as a means to identify its citizens for almost 20 years. During those years, several security issues have emerged. A new PhD dissertation from the University of Tartu has shed light on those issues and pointed […]

On Friday, August 28 Janno Siim defended his PhD thesis „Non-Interactive Shuffle Arguments„ A fundamental requirement for all democratic governments is a secure voting system. In recent years some countries, like Estonia and Switzerland, have adopted internet voting (i-voting), and many more have experimented (e.g., Norway and Australia) or have plans to adopt it in […]

On the 13th of August Karim Baghery defended his PhD thesis „Reducing trust and improving security in zk-SNARKs and commitments„ in which he studied the zk-SNARKs from two different perspectives, namely reducing trust in their setup phase and improving their security in different applications. On the need for proof systems People frequently are asked to […]