uvasrg / evademl Goto Github PK
View Code? Open in Web Editor NEWAn evolutionary framework for evading machine learning-based malware classifiers.
License: MIT License
An evolutionary framework for evading machine learning-based malware classifiers.
License: MIT License
Hi,
I have an issue in clverhans_models. I install cleverhans but I find there is no conv_2d in cleverhans.utils.
Do you have some special reference?
Thank you in advance.
Hi! I'm trying to replicate the results from your EvadeML GP attacks paper on PDFs. I ran into a couple of issues running the code. I trained the two PDF classifier models, and when trying to run Step 2 (with Step 1 detection server running) - ./utils/generate_ext_genome.py [classifier_name] [benign_sample_folder] [file_number]
for the two classifiers, I get two different errors:
PDFRate
$ ./utils/generate_ext_genome.py pdfrate ~/research/datasets/benign 1
Traceback (most recent call last):
File "./utils/generate_ext_genome.py", line 83, in <module>
selected_files = pdf_geno.select_files()
File "./utils/generate_ext_genome.py", line 38, in select_files
classifier_results = self.classifier(file_paths)
File "./utils/generate_ext_genome.py", line 24, in classifier
return self.classifier_func(*args)
File "./utils/generate_ext_genome.py", line 18, in <lambda>
self.classifier_func = lambda *args:query_classifier(classifier_name, *args)
File "/home/susobhan/research/EvadeML/utils/../lib/detector.py", line 22, in query_classifier
results = server.query_classifier(classifier_name, file_paths, seed_sha1)
File "/usr/lib/python2.7/xmlrpclib.py", line 1243, in __call__
return self.__send(self.__name, args)
File "/usr/lib/python2.7/xmlrpclib.py", line 1602, in __request
verbose=self.__verbose
File "/usr/lib/python2.7/xmlrpclib.py", line 1283, in request
return self.single_request(host, handler, request_body, verbose)
File "/usr/lib/python2.7/xmlrpclib.py", line 1316, in single_request
return self.parse_response(response)
File "/usr/lib/python2.7/xmlrpclib.py", line 1493, in parse_response
return u.close()
File "/usr/lib/python2.7/xmlrpclib.py", line 800, in close
raise Fault(**self._stack[0])
xmlrpclib.Fault: <Fault 1: "<type 'exceptions.ValueError'>:The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()">
Hidost -
$ ./utils/generate_ext_genome.py hidost ~/research/datasets/benign 1
Traceback (most recent call last):
File "./utils/generate_ext_genome.py", line 83, in <module>
selected_files = pdf_geno.select_files()
File "./utils/generate_ext_genome.py", line 38, in select_files
classifier_results = self.classifier(file_paths)
File "./utils/generate_ext_genome.py", line 24, in classifier
return self.classifier_func(*args)
File "./utils/generate_ext_genome.py", line 18, in <lambda>
self.classifier_func = lambda *args:query_classifier(classifier_name, *args)
File "/home/susobhan/research/EvadeML/utils/../lib/detector.py", line 22, in query_classifier
results = server.query_classifier(classifier_name, file_paths, seed_sha1)
File "/usr/lib/python2.7/xmlrpclib.py", line 1243, in __call__
return self.__send(self.__name, args)
File "/usr/lib/python2.7/xmlrpclib.py", line 1602, in __request
verbose=self.__verbose
File "/usr/lib/python2.7/xmlrpclib.py", line 1283, in request
return self.single_request(host, handler, request_body, verbose)
File "/usr/lib/python2.7/xmlrpclib.py", line 1316, in single_request
return self.parse_response(response)
File "/usr/lib/python2.7/xmlrpclib.py", line 1493, in parse_response
return u.close()
File "/usr/lib/python2.7/xmlrpclib.py", line 800, in close
raise Fault(**self._stack[0])
xmlrpclib.Fault: <Fault 1: "<type 'exceptions.AttributeError'>:'dict' object has no attribute 'decision_function'>
Could you please help me with this? Apologies if I'm doing something wrong here.
Hi there,
I am trying to generate a test case mentioned on Page 9 of your academic paper.
"However, the count javascript feature is not an accurate count of the number of embedded
JavaScript code pieces in a PDF. It just extracts the number of JavaScript keywords, but these keywords are optional in script execution. The targeted PDF reader will execute the JavaScript
even without the /Javascript keyword."
Can you help me with a test file or hash that can help me replicate this.
Hi! I am using hidost as a classifier for starting detection_agent_server. But classifiers/hidost_wrapper.py prompts <Fault 1: "<type ' exceptions.AttributeError'>:'dict' object has no attribute 'decision_function'">
The model file I am using is the SL2013.pickle
file generated in the repository hidost-reproduction (https://github.com/srndic/hidost-reproduction), what can I do to fix this?
It is listed in the help message that we can specify an oracle (like cuckoo in the case of your experiment in the research paper) using the "-o" argument but it doesn't seem to work? I also took a look at the gp.py script and there is no "-o" argument at all. Is this intended? If so, is there any way we can reproduce the experiment using cuckoo as an oracle?
Hi,
I am trying to replicate the experiment results. After following the instructions on the hidost-reproduction repo (downloading the data tar file and doing sudo make on root directory), I can't find the features.nppf file required for the project.conf entry in the main EvadeML root directory.
Am i required to do some work with the main hidost repo (https://github.com/srndic/hidost)? I was under the impression that the hidost-reproduction would completely reproduce the experiment and remove the need for the first repo.
Any help in understanding the situation will be helpful. Thank you!
I'm trying to reproduce your experiment and I'm having a few issues. If you have any insights into causes, I'd really appreciate any suggestions you have.
I have already installed the modified pdfrw (pdf malware parser). Any idea why this could be happening?
Could you provide the data used in your study to support replication and validation?
Specifically, providing a list of hashes for the seed malicious PDFs and the benign PDFs should allow your evaluation to be replicated with higher accuracy.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.