AI For Fake Artefact Generation / Detection
Details
Intro:
The scale and frequency with which information can be distributed on social media make the problem of fake news a rapidly metastasizing issue. To do any content filtering or labelling demands an algorithmic solution.
Recent advances in media generation techniques have made it
easier for attackers to create forged images and videos. State of-the-art methods enable the real-time creation of a forged version of a single video obtained from a social network. Although numerous methods have been developed for detecting forged images and videos, they are generally targeted at certain domains and quickly become obsolete as new kinds of attacks appear.
Agenda (lectures are in English):
- Amir Ivry, "Geometric Approach for Voice Activity Detection with Deep Learning"
This study addresses voice activity detection in acoustic environments of transients and stationary noises, which often occur in real life scenarios. Transient interferences can be affiliated with various fake artifacts that are manipulated to hurt the speech intelligibility and fidelity. Unique spatial patterns of speech and non-speech audio frames are exploited by independently learning their underlying geometric structure. This process is done through a deep encoder-decoder based neural network architecture. This structure involves an encoder that maps spectral features with temporal information to their low-dimensional representations, which are generated by applying the diffusion maps method. The encoder feeds a decoder that maps the embedded data back into the high-dimensional space. A deep neural network, which is trained to separate speech from nonspeech frames, is obtained by concatenating the decoder to the encoder. Experimental results show enhanced performance compared to competing voice activity detection methods. The improvement is achieved in both accuracy, robustness and generalization ability. Our model performs in a real-time manner and can be integrated into audio-based communication systems. A batch algorithm is also presented, which obtains an even higher accuracy for off-line applications.
Bio:
Amir Ivry received his BSc. in 2016 and currently pursues his PhD. (direct track), both in the Electrical Engineering Faculty in the Technion Israel Institute of Technology. His research focuses on deep learning for audio-based applications. Since 2015, Amir holds a position as a project leader and algorithms developer in the fields of machine learning, computer vision, image processing and signal processing in the Prime Minister’s Office.
- Dr. Eyal Gruss, "Fake Anything: "The Art of Deep Learning"
The age of creative machines is afoot. I will review recent state of the art applications of generative deep learning algorithms in image processing, language modeling and media arts. I will also exhibit my digital artwork and perform live demos of the new generation of deep learning algorithms. I will also present my latest work, "The Art of Deep Learning", which is a collection of short videos and slides that demonstrate the power of deep learning. [text partly generated by a neural network]
Bio:
Dr. Eyal Gruss is a machine learning researcher and consultant expertizing in image and language processing. Eyal Holds a PhD in physics and is a Talpiyot graduate. He is a mentor for Google Launchpad and gives lectures and workshops for both professionals and the general public. Eyal is also a new-media artist creating interactive installations and computer generated art. https://www.linkedin.com/in/eyalgruss
- Uri Eliabayev DeepFake.
בעידן בו טכנולוגיות חדשות מוסגלות ליצור פנים חדשות, לחקות אנושי או ליצור מצג שווא בקלות, חשוב לדון ולהכיר את ההיבטים האתיים של התופעה ולדעת כיצד להתמודד איתה.
בהרצאה הזו נסקור את הדרכים השונות להתמודד עם התופעה, נדון בפעולות שיש לנקוט כאנשי פיתוח כדי לצמצמה, נבין מה ההשלכות החוקיות והאתיות ולקינוח נעמוד על התמודדות של ממשלות וגופים בינלאומיים עם היבטים אלו.
Sponsors:
www.deeponcology.ai