Opens in a new tab<\/span><\/p>\n Frontiers in Artificial Intelligence<\/span> is a series of public lectures at Microsoft Research Cambridge featuring leading researchers in the field, focusing on the cutting edge topics at the intersection of machine learning, statistics, and artificial intelligence. Students, scientists, and engineers in academia and industry are all welcome to join us for these exciting talks and the opportunity to socialize with the Cam-bridge AI\/ML community.<\/p>\n Neural Code Comprehension: A Learnable Representation of Code Semantics<\/span><\/strong><\/p>\n Tal Ben-Nun<\/span><\/strong><\/p>\n In the era of \u201cBig Code\u201d, research is being conducted into automating the understanding of computer programs. Most of the current works base on techniques from Natural Language Processing and Deep Learning, which have been successful recently, attempting to process the code directly or using syntactic representations (e.g., ASTs and AST paths). However, to comprehend program semantics robustly, structural features of code have to be taken into account as well, including function calls, branching, and interchangeable order of statements. In this talk, I will present a novel processing technique to use Machine Learning for code semantics, and show how it applies to a variety of program analysis tasks. In particular, we stipulate that a robust distributional hypothesis of code applies to both human- and machine-generated programs. Following this hypothesis, we define an embedding space, inst2vec, based on an Intermediate Representation (IR) of the code that is independent of the source programming language. We provide a novel definition of contextual flow for this IR, leveraging both the underlying data- and control-flow of the program. We then analyze the embeddings quantitatively using analogies and clustering, and evaluate the learned representation on three different high-level tasks. We show that even without fine-tuning, a single Recurrent Neural Network (RNN) architecture and fixed inst2vec embeddings outperform specialized approaches for performance prediction (compute device mapping, optimal thread coarsening); and algorithm classification from raw code (104 classes), where we set a new state-of-the-art.<\/p>\n Opens in a new tab<\/span><\/p>\n Hoda Heidari – What can Fair ML learn from Economic Theories of Disruptive Justice? (opens in new tab)<\/span><\/a><\/p>\n Finale Doshi-Velez – Interpretability in Machine Learning: What it means, How we’re getting there. (opens in new tab)<\/span><\/a><\/p>\n Francis Bach\u00a0 – Optimal Algorithms for smooth and strongly convex distributed optimization in networks (opens in new tab)<\/span><\/a><\/p>\n Andreas Geiger – Probabilistic and Deep Models for 3D Reconstruction (opens in new tab)<\/span><\/a><\/p>\n Francesco Orabona – Coin Betting for Backprop without Learning rates and More<\/a><\/p>\n Regina Barzilay – How Can NLP Help Cure Cancer? (opens in new tab)<\/span><\/a><\/p>\n Aapo Hyvarinen – Nonlinear ICA using temporal structure: a principled framework for unsupervised deep learning\u00a0<\/a><\/p>\n Max Welling – Generalizing Convolutions for Deep Learning <\/a><\/p>\n Opens in a new tab<\/span><\/p>\n","protected":false},"excerpt":{"rendered":" Auditorium 21 Station Road Cambridge CB1 2FB View the whole series on talks.cam (opens in new tab) Opens in a new tab Frontiers in Artificial Intelligence is a series of public lectures at Microsoft Research Cambridge featuring leading researchers in the field, focusing on the cutting edge topics at the intersection of machine learning, […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr_startdate":"2019-02-28","msr_enddate":"2019-02-28","msr_location":"Microsoft Research Cambridge","msr_expirationdate":"","msr_event_recording_link":"","msr_event_link":"","msr_event_link_redirect":false,"msr_event_time":"13:00","msr_hide_region":false,"msr_private_event":false,"msr_hide_image_in_river":0,"footnotes":""},"research-area":[13556],"msr-region":[],"msr-event-type":[197944],"msr-video-type":[],"msr-locale":[268875],"msr-program-audience":[],"msr-post-option":[],"msr-impact-theme":[],"class_list":["post-494366","msr-event","type-msr-event","status-publish","hentry","msr-research-area-artificial-intelligence","msr-event-type-hosted-by-microsoft","msr-locale-en_us"],"msr_about":"\n\n Auditorium<\/p>\n 21 Station Road <\/p>\n View the whole series on talks.cam (opens in new tab)<\/span><\/a><\/p>\n Opens in a new tab<\/span><\/p>\n Frontiers in Artificial Intelligence<\/span> is a series of public lectures at Microsoft Research Cambridge featuring leading researchers in the field, focusing on the cutting edge topics at the intersection of machine learning, statistics, and artificial intelligence. Students, scientists, and engineers in academia and industry are all welcome to join us for these exciting talks and the opportunity to socialize with the Cam-bridge AI\/ML community.<\/p>\n Neural Code Comprehension: A Learnable Representation of Code Semantics<\/span><\/strong><\/p>\n Tal Ben-Nun<\/span><\/strong><\/p>\n In the era of \u201cBig Code\u201d, research is being conducted into automating the understanding of computer programs. Most of the current works base on techniques from Natural Language Processing and Deep Learning, which have been successful recently, attempting to process the code directly or using syntactic representations (e.g., ASTs and AST paths). However, to comprehend program semantics robustly, structural features of code have to be taken into account as well, including function calls, branching, and interchangeable order of statements. In this talk, I will present a novel processing technique to use Machine Learning for code semantics, and show how it applies to a variety of program analysis tasks. In particular, we stipulate that a robust distributional hypothesis of code applies to both human- and machine-generated programs. Following this hypothesis, we define an embedding space, inst2vec, based on an Intermediate Representation (IR) of the code that is independent of the source programming language. We provide a novel definition of contextual flow for this IR, leveraging both the underlying data- and control-flow of the program. We then analyze the embeddings quantitatively using analogies and clustering, and evaluate the learned representation on three different high-level tasks. We show that even without fine-tuning, a single Recurrent Neural Network (RNN) architecture and fixed inst2vec embeddings outperform specialized approaches for performance prediction (compute device mapping, optimal thread coarsening); and algorithm classification from raw code (104 classes), where we set a new state-of-the-art.<\/p>\n Opens in a new tab<\/span><\/p>\n Hoda Heidari – What can Fair ML learn from Economic Theories of Disruptive Justice?<\/a><\/p>\n Finale Doshi-Velez – Interpretability in Machine Learning: What it means, How we’re getting there.<\/a><\/p>\n Francis Bach\u00a0 – Optimal Algorithms for smooth and strongly convex distributed optimization in networks<\/a><\/p>\n Andreas Geiger – Probabilistic and Deep Models for 3D Reconstruction<\/a><\/p>\n Francesco Orabona – Coin Betting for Backprop without Learning rates and More<\/a><\/p>\n Regina Barzilay – How Can NLP Help Cure Cancer?<\/a><\/p>\n Aapo Hyvarinen – Nonlinear ICA using temporal structure: a principled framework for unsupervised deep learning\u00a0<\/a><\/p>\n Max Welling – Generalizing Convolutions for Deep Learning <\/a><\/p>\n Opens in a new tab<\/span><\/p>\n","tab-content":[{"id":0,"name":"About","content":"Frontiers in Artificial Intelligence<\/span> is a series of public lectures at Microsoft Research Cambridge featuring leading researchers in the field, focusing on the cutting edge topics at the intersection of machine learning, statistics, and artificial intelligence. Students, scientists, and engineers in academia and industry are all welcome to join us for these exciting talks and the opportunity to socialize with the Cam-bridge AI\/ML community.\r\n\r\n![\"\"](\"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-content\/uploads\/2018\/07\/tal-004-300x300.jpg\")
<\/span><\/strong><\/p>\n<\/h3>\n
\nCambridge
\nCB1 2FB<\/p>\n<\/span><\/strong><\/p>\n<\/h3>\n
<\/span><\/strong>\r\n\r\nNeural Code Comprehension: A Learnable Representation of Code Semantics<\/span><\/strong>\r\n\r\nTal Ben-Nun<\/span><\/strong>\r\n\r\nIn the era of \u201cBig Code\u201d, research is being conducted into automating the understanding of computer programs. Most of the current works base on techniques from Natural Language Processing and Deep Learning, which have been successful recently, attempting to process the code directly or using syntactic representations (e.g., ASTs and AST paths). However, to comprehend program semantics robustly, structural features of code have to be taken into account as well, including function calls, branching, and interchangeable order of statements. In this talk, I will present a novel processing technique to use Machine Learning for code semantics, and show how it applies to a variety of program analysis tasks. In particular, we stipulate that a robust distributional hypothesis of code applies to both human- and machine-generated programs. Following this hypothesis, we define an embedding space, inst2vec, based on an Intermediate Representation (IR) of the code that is independent of the source programming language. We provide a novel definition of contextual flow for this IR, leveraging both the underlying data- and control-flow of the program. We then analyze the embeddings quantitatively using analogies and clustering, and evaluate the learned representation on three different high-level tasks. We show that even without fine-tuning, a single Recurrent Neural Network (RNN) architecture and fixed inst2vec embeddings outperform specialized approaches for performance prediction (compute device mapping, optimal thread coarsening); and algorithm classification from raw code (104 classes), where we set a new state-of-the-art.\r\n<\/h3>"},{"id":1,"name":"Past Speakers","content":"Hoda Heidari - What can Fair ML learn from Economic Theories of Disruptive Justice?<\/a>\r\n\r\nFinale Doshi-Velez - Interpretability in Machine Learning: What it means, How we're getting there.<\/a>\r\n\r\nFrancis Bach\u00a0 - Optimal Algorithms for smooth and strongly convex distributed optimization in networks<\/a>\r\n\r\nAndreas Geiger - Probabilistic and Deep Models for 3D Reconstruction<\/a>\r\n\r\nFrancesco Orabona - Coin Betting for Backprop without Learning rates and More<\/a>\r\n\r\nRegina Barzilay - How Can NLP Help Cure Cancer?<\/a>\r\n\r\nAapo Hyvarinen - Nonlinear ICA using temporal structure: a principled framework for unsupervised deep learning\u00a0<\/a>\r\n\r\nMax Welling - Generalizing Convolutions for Deep Learning <\/a>\r\n\r\n "}],"msr_startdate":"2019-02-28","msr_enddate":"2019-02-28","msr_event_time":"13:00","msr_location":"Microsoft Research Cambridge","msr_event_link":"","msr_event_recording_link":"","msr_startdate_formatted":"February 28, 2019","msr_register_text":"Watch now","msr_cta_link":"","msr_cta_text":"","msr_cta_bi_name":"","featured_image_thumbnail":null,"event_excerpt":"Frontiers in Artificial Intelligence is a series of public lectures at Microsoft Research Cambridge featuring leading researchers in the field, focusing on the cutting edge topics at the intersection of machine learning, statistics, and artificial intelligence. Students, scientists, and engineers in academia and industry are all welcome to join us for these exciting talks and the opportunity to socialize with the Cam-bridge AI\/ML community. Neural Code Comprehension: A Learnable Representation of Code Semantics Tal Ben-Nun…","msr_research_lab":[199561],"related-researchers":[],"msr_impact_theme":[],"related-academic-programs":[],"related-groups":[],"related-projects":[],"related-opportunities":[],"related-publications":[],"related-videos":[],"related-posts":[],"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/494366","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-event"}],"version-history":[{"count":6,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/494366\/revisions"}],"predecessor-version":[{"id":1147092,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/494366\/revisions\/1147092"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=494366"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=494366"},{"taxonomy":"msr-region","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-region?post=494366"},{"taxonomy":"msr-event-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event-type?post=494366"},{"taxonomy":"msr-video-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-video-type?post=494366"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=494366"},{"taxonomy":"msr-program-audience","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-program-audience?post=494366"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=494366"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=494366"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}