RT @EdwardRaffML: Lead by @rjjoyce8 , #EMBER24 has arrived @kdd_news #KDD25, the best, most open, and versatile malware detection benchmark….

RT @FSFG: EMBER2024 -- A Benchmark Dataset for Holistic Evaluation of Malware Classifiers.

RT @arxivsanitybot: Researchers unveil EMBER2024, a groundbreaking dataset advancing malware analysis. With diverse….

RT @reevesrs24: A new EMBER malware benchmark has been released, with new datasets and 14 baseline LightGBM models. It’s one of the best op….

@mrphilroth We trained 14 different classifiers, one for each of the file types, labels, and tags (plus a combined model). Check out our GitHub repo, which makes it easy to train and evaluate your own models on EMBER2024 and compare them to our benchmarks.

We have also made some major updates to the EMBER feature vector format. @mrphilroth re-implemented the entire package to use pefile instead of LEIF (which was pinned to Python 3.6), and we've added DOS header, Rich header, Authenticode signature, and PE parse warning features.

EMBER2024 also includes a "challenge" set - 6,315 files that went fully undetected on VirusTotal on first upload, but were later found to be malicious. This is a great resource for testing whether your classifier can handle evasive malware.

EMBER2024 has 6 different file types and 7 different types of labels and tags supporting tasks like malicious/benign detection, family classification, and behavior identification. The dataset is chunked evenly by week so that researchers can explore malware trends over time.

I'm excited to announce the EMBER2024 dataset, created in collaboration with @mrphilroth, @drhyrum, @EdwardRaffML, @rjzak, and others. The dataset brings an update of 3.2 million malicious and benign files first seen between Sep. 2023 and Dec. 2024.

To label your own malware collection, download ClarAVy here!.

RT @EdwardRaffML: Helping analysts with their reverse engineering work and finding similar functions is a problem. A lot of existing litera….

RT @EdwardRaffML: I love this for bitcoin belivers.

RT @umbccsee: 15 UMBC CSEE students successfully defended their PhD dissertations this year so far, bringing the total number of PhDs the d….

RT @EdwardRaffML: 𝑨𝒔𝒔𝒆𝒎𝒃𝒍𝒂𝒈𝒆: 𝑨𝒖𝒕𝒐𝒎𝒂𝒕𝒊𝒄 𝑩𝒊𝒏𝒂𝒓𝒚 𝑫𝒂𝒕𝒂𝒔𝒆𝒕 𝑪𝒐𝒏𝒔𝒕𝒓𝒖𝒄𝒕𝒊𝒐𝒏 𝒇𝒐𝒓 𝑴𝒂𝒄𝒉𝒊𝒏𝒆 𝑳𝒆𝒂𝒓𝒏𝒊𝒏𝒈 Compiling projects @github . Randomize the compile….

RT @EdwardRaffML: Malware Bytes, an article going into some of the big-picture factors on the malware work I've done over the past ~8 years….

RT @EdwardRaffML: Inspired by @_albertgu recent works in state space models, can we merge them with #VSA and #HRR for our long sequence cla….

RT @rharang: I've recommended this to lots of people who are in security and want to dive deeper into the details of ML, and I always hear….

Very excited to be in Copenhagen for @acm_ccs! I'll be presenting our research on embedding VirusTotal scans (in collaboration w/ @EdwardRaffML and @cknicholas) at the AISec workshop tomorrow.

RT @CyberRaiju: It took me almost 3 years, but I finally restructured the Practical Malware Analysis section of my site. This is now broken….

RT @EdwardRaffML: When I started teaching, I was surprised at how little research there is to what impacts students grades. Thanks to @UMBC….