The present work aims to reproduce and replicate the results of a prominent study in the computational neuroscience field. The objectives of this reproduction are: verifying the original claims and results as well as assessing the degree of replicability of the research. Initially, an introduction to the concepts of Reproduction and Replication on the computational science field is provided, along with their importance and the most common challenges that researchers face. Next, a brief overview of the current scenery in Computational Neuroscience is provided, and the selected work - 'The Microcircuits of Striatum in Silico' is presented. In the original study, the authors have presented a detailed neural network model of the striatum, a nucleus in the subcortical basal ganglia of the forebrain. The methodology employed for the replication focuses on two scopes: the influence of the network size and of the model complexity on the observed dynamics of the neural circuit. Different combinations of both scale and complexity reduction have been performed and compared. The results show a similar qualitative behaviour of the replicated models, considering the limitations and simplifications adopted. Although some key documentations on the original simulation are unavailable, hindering the comprehension and exact replication of the work, there was a solid effort of the authors of the original study to provide the information required to utilize their tool. Lastly, some recommendations are made for future works: improvements that can be implemented on the replicated models, and evaluation and feedback on the documentation and accessibility of the original research.
Questo lavoro ha l'obbiettivo di riprodurre e replicare i risultati di un importante studio nel campo delle neuroscienze computazionali. Gli obiettivi di questa tesi sono: verificare i risultati originali, così come valutare il grado di replicabilità della ricerca. Inizialmente abbiamo presentato una introduzione ai concetti di replicabilita' e riproducibilita' nel campo delle neuroscienze computazionali, sottolineando la loro importanza e quale sono le sfide più comuni. Poi, abbiamo fatto una breve panoramica dello scenario attuale e dello studio selezionato - 'The Microcircuits of Striatum in Silico'-, dove gli autori hanno presentato un modello di rete neurale molto dettagliato dello striato, un nucleo subcorticale dei gangli della base. La metodologia impiegata nella replicazione si concentra su due ambiti: l'influenza sulla dinamica della rete neurale in oggetto della dimensione della rete e della complessità del modello usato. Diverse combinazione di scala e complessità sono state simulate e quindi comparate. I risultati ottenuti mostrano un comportamento qualitativamente simile nei modelli replicati, specialmente considerando le limitazioni trovate e le semplificazioni adottate. Sebbene qualche documentazione chiave sulla simulazione originale non fosse disponibile, rendendo più difficile la comprensione e replica esatta del lavoro, c'è stato un solido sforzo dei autori dello studio originale per fornire le informazioni necessarie per utilizzare lo strumento da loro sviluppato. Infine, abbiamo proposto alcune raccomandazioni per i lavori futuri: i miglioramenti che possono essere implementati nei modelli, e una valutazione e feedback sulla documentazione e l'accessibilità della ricerca originale.
Replication of a computational model of the striatum with a multiscale approach
FURQUIM DAUD, PAOLLA
2021/2022
Abstract
The present work aims to reproduce and replicate the results of a prominent study in the computational neuroscience field. The objectives of this reproduction are: verifying the original claims and results as well as assessing the degree of replicability of the research. Initially, an introduction to the concepts of Reproduction and Replication on the computational science field is provided, along with their importance and the most common challenges that researchers face. Next, a brief overview of the current scenery in Computational Neuroscience is provided, and the selected work - 'The Microcircuits of Striatum in Silico' is presented. In the original study, the authors have presented a detailed neural network model of the striatum, a nucleus in the subcortical basal ganglia of the forebrain. The methodology employed for the replication focuses on two scopes: the influence of the network size and of the model complexity on the observed dynamics of the neural circuit. Different combinations of both scale and complexity reduction have been performed and compared. The results show a similar qualitative behaviour of the replicated models, considering the limitations and simplifications adopted. Although some key documentations on the original simulation are unavailable, hindering the comprehension and exact replication of the work, there was a solid effort of the authors of the original study to provide the information required to utilize their tool. Lastly, some recommendations are made for future works: improvements that can be implemented on the replicated models, and evaluation and feedback on the documentation and accessibility of the original research.File | Dimensione | Formato | |
---|---|---|---|
Master_Thesis_Paolla_Furquim.pdf
embargo fino al 08/12/2024
Dimensione
33.7 MB
Formato
Adobe PDF
|
33.7 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/183869