{"id":13,"date":"2016-03-23T10:02:53","date_gmt":"2016-03-23T09:02:53","guid":{"rendered":"http:\/\/pages.di.unipi.it\/dbacciu\/?page_id=13"},"modified":"2024-01-03T13:44:08","modified_gmt":"2024-01-03T12:44:08","slug":"publications","status":"publish","type":"page","link":"https:\/\/pages.di.unipi.it\/bacciu\/publications\/","title":{"rendered":"Publications"},"content":{"rendered":"\n

Here you can find a consolidated (a.k.a. slowly updated) list of my publications. A frequently updated (and possibly noisy) list of works is available on my Google Scholar profile<\/a>.<\/p>\n\n\n\n

Please find below a short list of highlight publications for my recent activity.<\/p>\n\n\n

<\/a><\/form>
\"Dual<\/div>

Numeroso, Danilo; Bacciu, Davide; Veli\u010dkovi\u0107, Petar<\/p>

Dual Algorithmic Reasoning<\/a> Conference<\/span> <\/p>

Proceedings of the Eleventh International Conference on Learning Representations (ICLR 2023), <\/span>2023<\/span>, (Notable Spotlight paper)<\/span>.<\/p>

Abstract<\/a><\/span> | Links<\/a><\/span> | BibTeX<\/a><\/span><\/p>

@conference{Numeroso2023,
\r\ntitle = {Dual Algorithmic Reasoning},
\r\nauthor = {Danilo Numeroso and Davide Bacciu and Petar Veli\u010dkovi\u0107},
\r\nurl = {https:\/\/openreview.net\/pdf?id=hhvkdRdWt1F},
\r\nyear  = {2023},
\r\ndate = {2023-05-01},
\r\nurldate = {2023-05-01},
\r\nbooktitle = {Proceedings of the Eleventh International Conference on Learning Representations (ICLR 2023)},
\r\nabstract = {Neural Algorithmic Reasoning is an emerging area of machine learning which seeks to infuse algorithmic computation in neural networks, typically by training neural models to approximate steps of classical algorithms. In this context, much of the current work has focused on learning reachability and shortest path graph algorithms, showing that joint learning on similar algorithms is beneficial for generalisation. However, when targeting more complex problems, such \"similar\" algorithms become more difficult to find. Here, we propose to learn algorithms by exploiting duality of the underlying algorithmic problem. Many algorithms solve optimisation problems. We demonstrate that simultaneously learning the dual definition of these optimisation problems in algorithmic learning allows for better learning and qualitatively better solutions. Specifically, we exploit the max-flow min-cut theorem to simultaneously learn these two algorithms over synthetically generated graphs, demonstrating the effectiveness of the proposed approach. We then validate the real-world utility of our dual algorithmic reasoner by deploying it on a challenging brain vessel classification task, which likely depends on the vessels\u2019 flow properties. We demonstrate a clear performance gain when using our model within such a context, and empirically show that learning the max-flow and min-cut algorithms together is critical for achieving such a result.},
\r\nnote = {Notable Spotlight paper},
\r\nkeywords = {},
\r\npubstate = {published},
\r\ntppubtype = {conference}
\r\n}
\r\n<\/pre><\/div>
Close<\/a><\/p><\/div>
Neural Algorithmic Reasoning is an emerging area of machine learning which seeks to infuse algorithmic computation in neural networks, typically by training neural models to approximate steps of classical algorithms. In this context, much of the current work has focused on learning reachability and shortest path graph algorithms, showing that joint learning on similar algorithms is beneficial for generalisation. However, when targeting more complex problems, such "similar" algorithms become more difficult to find. Here, we propose to learn algorithms by exploiting duality of the underlying algorithmic problem. Many algorithms solve optimisation problems. We demonstrate that simultaneously learning the dual definition of these optimisation problems in algorithmic learning allows for better learning and qualitatively better solutions. Specifically, we exploit the max-flow min-cut theorem to simultaneously learn these two algorithms over synthetically generated graphs, demonstrating the effectiveness of the proposed approach. We then validate the real-world utility of our dual algorithmic reasoner by deploying it on a challenging brain vessel classification task, which likely depends on the vessels\u2019 flow properties. We demonstrate a clear performance gain when using our model within such a context, and empirically show that learning the max-flow and min-cut algorithms together is critical for achieving such a result.<\/div>
Close<\/a><\/p><\/div>