Bruno Grisci

📢 New Paper Out! I'm happy to share that our latest research, "Assessing Feature Scorer Results on High-Dimensional Datasets with t-SNE", has just been accepted for publication in Neurocomputing 📈 Visualizations: sbcblab.github.io/wtsne/ 📄 Paper: doi.org/10.1016/j.neuc…

My thanks to everyone involved in the project and in the development of this solution, It is meaningful to see an initiative of this kind receive this level of recognition.

This recognition also reinforces the value of applied research and collaboration between academia and industry. Results like this depend on technical competence, interdisciplinary perspective, and, above all, teamwork.

The granting of this patent represents a very relevant outcome of work developed during an earlier stage of my academic and professional trajectory, when I was at @Unisinos collaborating with @DellnoBrasil .

Importantly, scorers that appear similar numerically can produce very different effects when visualized, a reminder that context matters. #MachineLearning #DataScience #FeatureSelection #Interpretability #DataVisualization #tSNE #Neurocomputing #AIResearch

In some cases, even simple statistical scorers outperformed complex models It can reveal outliers and help detect mislabeled samples. Visualization quality (measured by silhouette and trustworthiness) correlates with downstream classification performance.

📢 New Paper Out! I'm happy to share that our latest research, "Assessing Feature Scorer Results on High-Dimensional Datasets with t-SNE", has just been accepted for publication in Neurocomputing 📈 Visualizations: sbcblab.github.io/wtsne/ 📄 Paper: doi.org/10.1016/j.neuc…

X apagou postagens preconceituosas e com manifestações de ódio de seu chatbot. Atualização havia permitido comentários ‘politicamente incorretos’. nexojornal.com.br/expresso/2025/…

🎉"Os algoritmos de otimização evolutiva moldam o aprendizado e o comportamento de redes neurais artificiais de forma diferente dos métodos tradicionais?", de Bruno Grisci, é contemplado pelo Instituto Serrapilheira! 🌐🔗Leia mais em: inf.ufrgs.br/site/?post_typ… #INF #UFRGS

Uma boa notícia para a ciência: o Serrapilheira acaba de selecionar 16 jovens cientistas para receberem apoio financeiro de R$ 7,4 milhões para desenvolverem suas pesquisas. Confira a lista! serrapilheira.org/aprovados-em-n…

Dois pesquisadores da UFRGS são contemplados em edital nacional do Instituto Serrapilheira - Projetos selecionados investigam envelhecimento cerebral (Wyllians Borelli) e novas formas de aprendizado em redes neurais artificiais (Bruno Iochins): tinyurl.com/serraufrgs

@danielreen @_Scarlat @Night_WalkerXD

#SORTEIO🎉 Em gratidão ao apoio de vocês ao meu trabalho, vou sortear um Switch 2 aos membros do ofertasnintendo.com.br - Me segue aí - 💙Curtir +🔄 RT sem comentar - Marque 2 amigos com quem jogaria um Mario Kart e coloque o print de qual grupo você faz parte. Sorteio em 23/06

it was the best of timezones, it was the worst of timezones

This might be the best visualization of limitations of LLMs. LLms learn from a distribution. The further away from this distribution we ask the LLM, the less reliable the answer. Larger LLMs = better performance, but inherently it's the same pattern.LLM's