Search

Edge Computing and Machine Learning have become increasingly vital in today’s digital landscape. Edge computing brings computational power closer to the data source enabling reduced latency and bandwith, increased privacy, and real-time decision-making. Running Machine Learning models on edge devices further enhances these advantages by reducing the reliance on cloud. This empowers industries such as transport, healthcare, manufacturing, to harness the full potential of Machine Learning. MLOps, or Machine Learning Operations play a major role streamlining the deployment, monitoring, and management of Machine Learning models in production. With MLOps, organisations can achieve faster model iteration, reduced deployment time, improved collaboration with developers, optimised performance, and ultimately meaningful business outcomes. Integrating MLOps with edge devices poses unique challenges. Overcoming these challenges requires careful planning, customised deployment strategies, and efficient model optimization techniques. This thesis project introduces a set of tools that enable the integration of MLOps practices with edge devices. The solution consists of two sets of tools: one for setting up infrastructure within edge devices to be able to receive, monitor, and run inference on Machine Learning models, and another for MLOps pipelines to package models to be compatible with the inference and monitoring components of the respective edge devices. This platform was evaluated by obtaining a public dataset used for predicting the breakdown of Air Pressure Systems in trucks, which is an ideal use-case for running ML inference on the edge, and connecting MLOps pipelines with edge devices.. A simulation was created using the data in order to control the volume of data flow into edge devices. Thereafter, the performance of the platform was tested against the scenario created by the simulation script. Response time and CPU usage in different components were the metrics that were tested. Additionally, the platform was evaluated against a set of commercial and open source tools and services that serve similar purposes. The overall performance of this solution matches that of already existing tools and services, while allowing end users setting up Edge-MLOps infrastructure the complete freedom to set up their system without completely relying on third party licensed software.MLOps-integraatio reunalaskennan tarpeisiin. Tiivistelmä. Reunalaskennasta (Edge Computing) ja koneoppimisesta on tullut yhä tärkeämpiä nykypäivän digitaalisessa ympäristössä. Reunalaskenta tuo laskentatehon lähemmäs datalähdettä, mikä mahdollistaa reaaliaikaisen päätöksenteon ja pienemmän viiveen. Koneoppimismallien suorittaminen reunalaitteissa parantaa näitä etuja entisestään vähentämällä riippuvuutta pilvipalveluista. Näin esimerkiksi liikenne-, terveydenhuolto- ja valmistusteollisuus voivat hyödyntää koneoppimisen koko potentiaalin. MLOps eli Machine Learning Operations on merkittävässä asemassa tehostettaessa ML -mallien käyttöönottoa, seurantaa ja hallintaa tuotannossa. MLOpsin avulla organisaatiot voivat nopeuttaa mallien iterointia, lyhentää käyttöönottoaikaa, parantaa yhteistyötä kehittäjien kesken, optimoida laskennan suorituskykyä ja lopulta saavuttaa merkityksellisiä liiketoimintatuloksia. MLOpsin integroiminen reunalaitteisiin asettaa ainutlaatuisia haasteita. Näiden haasteiden voittaminen edellyttää huolellista suunnittelua, räätälöityjä käyttöönottostrategioita ja tehokkaita mallien optimointitekniikoita. Tässä opinnäytetyöhankkeessa esitellään joukko työkaluja, jotka mahdollistavat MLOps-käytäntöjen integroinnin reunalaitteisiin. Ratkaisu koostuu kahdesta työkalukokonaisuudesta: toinen infrastruktuurin perustamisesta reunalaitteisiin, jotta ne voivat vastaanottaa, valvoa ja suorittaa päätelmiä koneoppimismalleista, ja toinen MLOps “prosesseista”, joilla mallit paketoidaan yhteensopiviksi vastaavien reunalaitteiden komponenttien kanssa. Ratkaisun toimivuutta arvioitiin avoimeen dataan perustuvalla käyttötapauksella. Datan avulla luotiin simulaatio, jonka tarkoituksena oli mahdollistaa reunalaitteisiin suuntautuvan datatovirran kontrollonti. Tämän jälkeen suorituskykyä testattiin simuloinnin luoman skenaarion avulla. Testattaviin mittareihin kuuluivat muun muassa suorittimen käyttö. Lisäksi ratkaisua arvioitiin vertaamalla sitä olemassa oleviin kaupallisiin ja avoimen lähdekoodin alustoihin. Tämän ratkaisun kokonaissuorituskyky vastaa jo markkinoilla olevien työkalujen ja palvelujen suorituskykyä. Ratkaisu antaa samalla loppukäyttäjille mahdollisuuden perustaa Edge-MLOps-infrastruktuuri ilman riippuvuutta kolmannen osapuolen lisensoiduista ohjelmistoista.

Pink Miana plant (Coleus scuttelarioides) is a miana variety that is only cultivated in one of the tourism msmes The Lawu Fresh, Puntukrejo Ngargoyoso Village, Karanganyar, Central Java. This ornamental plant, which is famous for its unique leaf color, turns out to contain various substances such as antioxidants and anti-inflammatories that are good for the body's immunity. However, the abundance of potential possessed by Miana is not balanced with the capabilities of the local community in its processing. Improper processing of Pink Miana can lead to reduced or loss of content of these substances if processed with too hot temperatures. The method used in solving these problems is controlled boiling. Controlled boiling is a method of boiling using low heat with regular temperature monitoring so as not to exceed the maximum temperature that the material can accept. The findings obtained are syrup products made from the first miana leaves "Pink Fresh" which contains high antioxidants and anti-inflammatories that are efficacious for the body's immunity. The purpose of making functional drinks based on miana is to provide an innovative variety of health drinks compared to those already on the market. The creation of functional beverage products based on miana "Pink Fresh" is expected to be a product supporting the body's immune system by implementing learning-based activities to utilize the potential of the village into a product with high economic value in order to achieve the economic empowerment of the village community. Young people are the main target market of the "Pink Fresh" product because of its taste innovation that is unlike other health drinks. Despite using a pre-order system, the first sales of the "Pink Fresh" product received rave reviews from consumers until it reached 70 bottles in the first order. Abstrak Tanaman Pink Miana (Coleus scuttelarioides) merupakan varietas miana yang hanya dibudidayakan di salah satu UMKM pariwisata The Lawu Fresh, Desa Puntukrejo Ngargoyoso, Karanganyar, Jawa Tengah. Tanaman hias yang terkenal dengan keunikan warna daunnya ini ternyata mengandung berbagai zat seperti antioksidan serta antiinflamasi yang baik bagi imunitas tubuh. Akan tetapi, melimpahnya potensi yang dimiliki miana tidak diimbangi dengan kapabilitas masyarakat setempat dalam pengolahannya. Pengolahan Pink Miana yang kurang tepat dapat menyebabkan berkurang atau hilangnya kandungan zat- zat tersebut jika diolah dengan suhu yang terlalu panas. Metode yang digunakan dalam pemecahan permasalahan tersebut adalah controlled boiling. Controlled boiling merupakan metode merebus menggunakan api kecil dengan pengawasan suhu yang teratur agar tidak melampaui suhu maksimal yang dapat diterima bahan. Adapun hasil temuan yang diperoleh adalah produk sirup berbahan dasar daun miana pertama “Pink Fresh” yang mengandung antioksidan serta antiinflamasi tinggi yang berkhasiat bagi imunitas tubuh. Tujuan dari pembuatan minuman fungsional berbahan dasar miana adalah memberikan variasi minuman kesehatan yang inovatif dibandingkan yang telah ada di pasaran. Terciptanya produk minuman fungsional berbahan dasar miana “Pink Fresh” diharapkan dapat menjadi produk penunjang imun tubuh dengan penerapan kegiatan berbasis based learning untuk memanfaatkan potensi desa menjadi produk dengan nilai ekonomis tinggi guna mencapai keberdayaan ekonomi masyarakat desa. Kalangan muda menjadi target pasar utama dari produk “Pink Fresh” karena inovasi rasanya yang tidak seperti minuman kesehatan lainnya. Meskipun menggunakan sistem pre-order, penjualan pertama dari produk “Pink Fresh” mendapat sambutan hangat dari para konsumen hingga mencapai 70 botol dalam pesanan pertama.

Industry 4.0 constituted a trigger to a new phase in the Industrial Revolution, heavily focused on the interconnectivity of the systems, bringing disruptive technologies such as Additive Manufacturing (AM). On top of that, in the past few decades AM has been gaining visibility due to technological evolution and due to the need to increase differentiating factors to overtake competitors. Coupled with this, phenomenon such as increased firms’ competitivity and rivalry, increased customer demand and diversity and the shift from an industrial economy to a knowledge-based economy are leading to the emergence of new business models. However, despite of AM's business possibilities, in literature it is missing a model to analyse the impact of AM in business strategy. Accordingly, the present research focuses on understanding how AM impacts firms’ business strategy. To achieve that objective, is proposed an adaptation of Porter's Five Forces Model to aid in positioning firms’ strategy when using AM. The model was applied to 13 organisations operating with AM in Portugal. It was possible to conclude that all the participating firms, except one, fit in the incremental stream of development as regards to AM technology, where AM technologies appear as a complementing technology. Also, the participating firms seem to experience the low capability to capture specialized workforce for AM, high capital requirements to enter the market and low IPR regulation. AM brings an opportunity for higher bargaining power to arise due to 'prosumerism', yet it does not add value as a tool for the standard products industry. Moreover, suppliers strongly influence sectors' competition, which will presumably suffer from increased rivalry tensions. info:eu-repo/semantics/publishedVersion

Antimicrobial peptides (AMPs) have shown cell membrane-directed mechanisms of action. This specificity can be effective against infectious agents that have acquired resistance to conventional drugs. The AMPs’ membrane-specificity and their great potential to combat resistant microbes has brought hope to the medical/therapeutic scene. The high death rate worldwide due to antimicrobial resistance (AMR) has pushed forward the search for new molecules and product developments, mainly antibiotics. In the current scenario, other strategies including the association of two or more drugs have contributed to the treatment of difficult-to-treat infectious diseases, above all, those caused by bacteria. In this context, the synergistic action of AMPs associated with current antibiotic therapy can bring important results for the production of new and effective drugs to overcome AMR. This review presents the advances obtained in the last 5 years in medical/antibiotic therapy, with the use of products based on AMPs, as well as perspectives on the potentialized effects of current drugs combined with AMPs for the treatment of bacterial infectious diseases.