הפקולטה להנדסה ע"ש איבי ואלדר פליישמן

EE Seminar: Polynomial Time Cryptanalytic Extraction of Neural Network Models

09 בדצמבר 2024, 12:00

אולם 011, בניין כיתות-חשמל

EE Seminar: Polynomial Time Cryptanalytic Extraction of Neural Network Models

(The talk will be given in English)

Speaker: Prof. Adi Shamir

Computer Science and Applied Mathematics,Weizmann Institute of Science

011 hall, Electrical Engineering-Kitot Building‏

Monday, December 9^th, 2024

12:00 - 13:00

Polynomial Time Cryptanalytic Extraction of Neural Network Models

Abstract

Abstract. Billions of dollars and countless GPU hours are currently spent on training Deep Neural Networks (DNNs) for a variety of tasks. Thus, it is essential to determine the difficulty of extracting all the parameters of such neural networks when given access to their black-box implementations. Many versions of this problem have been studied over the last 30 years, and the best previous attack on ReLU-based deep neural networks was presented at Crypto’20 by Carlini, Jagielski, and Mironov. It resembled a differential chosen plaintext attack on a cryptosystem, which has a secret key embedded in its black-box implementation and required a polynomial number of queries but an exponential amount of time (as a function of the number of neurons).

In this talk, I will improve this attack by developing several new techniques that make it possible to extract with arbitrarily high precision all the real-valued parameters of a ReLU-based DNN using a polynomial number of queries AND a polynomial amount of time. We demonstrated its practical efficiency by applying it to a full-sized neural network for classifying the CIFAR10 dataset, which has 3072 inputs, 8 hidden layers with 256 neurons each, and about 1.2 million neuronal parameters. An attack following the approach by Carlini et al. requires an exhaustive search over 2^256 possibilities, whereas our attack requires only 30 minutes on a 256-core computer. In the last part of the talk I will show a recent extension of the attack which can extract in polynomial time all the parameters of the network even when the attacker receives only the label (e.g., a “cat” or a “dog”) of any given input, rather than the numeric values of its logits.

השתתפות בסמינר תיתן קרדיט שמיעה = עפ"י רישום שם מלא + מספר ת.ז. בטופס הנוכחות שיועבר באולם במהלך הסמינר

ברכות חמות לפרופ' ראמי חג'-עלי על זכייתו במענק המשותף MOONSHOT-MED של אוניברסיטת תל אביב ושירותי בריאות כללית

ראמי, מחקר שתורם משמעותית לקידום הידע בתחום ומוביל לתוצאות פורצות דרך.

הצלחה רבה בהמשך דרכו

TAUSAT1

TAUSAT2

TAUSAT3
Tevel1

Tevel2

Optical communications

Hyperspectral remote sensing

תכנית ת.ב.ל- תלמידים בונים לוויינים

תכנית לאומית לבנייה ושיגור להק לוויינים על ידי תלמידי תיכון מכל רחבי הארץ!

על התכנית:

תבל הינה תכנית מדעית חינוכית של סוכנות החלל הישראלית ואוניברסיטת תל אביב במסגרתה תלמידי תיכון בונים לווייני מחקר זעירים ומשגרים אותם לחלל. התכנית נועדה לחשוף את התלמידים לתחום של הנדסת חלל ולאפשר להם להתנסות בפרויקט הנדסי אמיתי. במהלך הפרויקט התלמידים עוברים את כל שלבי המחקר והפיתוח החל מתכנון ראשוני של הלוויין, הגדרת המשימה המדעית ודרישות המערכת ועד לפיתוח התוכנה המוטסת, בדיקות מוכנות לשיגור ותפעול הלוויינים בחלל.

בתכנית משתתפים תלמידים מ-9 רשויות: שער הנגב, מעלה אדומים, גבעת שמואל, טייבה, כפר קרע, עין מאהל, ירכא, ירוחם והרצליה. בכל קבוצה נבנה לוויין בחדר הנקי שהוקם בבית הספר. יחד ישוגרו כל 9 הלוויינים לחלל ויבצעו מחקר מדעי משותף.

הלמידה בתכנית כוללת למידה תיאורטית של נושאים מדעיים וטכנולוגיים בהם פיזיקה, אלקטרוניקה, תוכנה, הנדסה מכאנית, תקשורת רדיו, לצד התנסות מעשית (hands-on) בפיתוח הלוויינים, כתיבת תוכנה מוטסת, בדיקות מערכת, תפעול תחנת תקשורת לוויינים, תכנון מכאני ותרמי, תכנון מסלולים ועוד. התלמידים מתמודדים עם אתגרים אמיתיים הדורשים מהם יכולת למידה והכוונה עצמית, חשיבה ביקורתית, חשיבה יצירתית, התנסות בתהליכי פתרון בעיות ועבודת צוות.

(לשים תמונה של לתמידים בכיתה- ילדים עובדים יחד)

על הלוויינים:

בתכנית מפותחים 9 ננו לוויינים מסוג לווייני קובייה (cubesat) בגודל U1- 10*10*11.3 ס"מ

משימות מדעיות לווייני תבל2: ללווייני תבל2 שתי משימות מדעיות:

מדידת קרינה מהחלל: מטע"ד המפותח ע"י מחלקת סביבת חלל במרכז למחקר גרעיני שורק (ממ"ג) ועתיד לבצע ניסוי למדידת שטף חלקיקים אנרגטיים הנפלטים מהשמש ומקרניים קוסמיות. המדידה תבוצע ע"י ספירת מספר היפוכי הביט, תיעוד קצרים וקרינה מצטברת הנמדדת ברכיבי האלקטרוניקה והחיישנים שבמטע"ד.
תקשורת חובבות רדיו באמצעות משיב אנלוגי

שיגור: מרץ 2025, SpaceX

מרכז בקרה ושליטה: הלוויינים נשלטים מתוך תחנת הקרקע לתקשורת לוויינים באוניברסיטת תל אביב.

A group of young men looking at a computer screen

Description automatically generated A group of people sitting in chairs using computers

Description automatically generated

סמינר שמיעה לתלמידי תואר שני ושלישי

03 בדצמבר 2024, 14:00

בניין וולפסון, חדר 206

Join us with Zoom

Involving humans in AI-supported fake news detection

Amit Miller, M.Sc. student at the department of Industrial Advisor: Prof. Joachim Meyer

Abstract:

“Fake News” refers to false or misleading information presented as news, often spread online or through social media. It is typically created to manipulate public opinion, generate revenue through sensationalism, or promote specific agendas .An overwhelming amount of fake news is circulating online, making it increasingly difficult to distinguish truth from falsehood. This vast spread creates confusion, as fake news often mimics credible sources, blurring the line between real and fabricated information. Our research analyzes a system, such as a news desk at a media site, that receives a stream of incoming news items from various sources and has to decide which items to publish and which to discard as fake. To do so, one or more people can view items and there is also an AI system that classifies the incoming items. We evaluated various configurations within this hybrid setup with a computational model. We reviewed performance metrics in order to provide recommendations on the use of such a system as a function of properties of the AI and the human, the costs and benefits of publishing or not publishing fake or true items, and the relative frequency of fake items in the incoming stream. The results show that a clear recommendation can be given in most cases, and that some configurations are superior to others in the cases we reviewed. Additionally, we were able to show that some parameters can be combined to reduce the complexity of the problem.

Bio:

Amit Miller is a M.Sc. student in the department of industrial engineering in Tel Aviv university. He works as a professional in the gaming industry, using data, analytics & AI in order to optimize business decisions. He also holds a B.Sc. degree in Civil engineering from the Technion.

ד"ר גלי פיכמן-Bio inspired Design of Peptide-Baseded Infections Functional Gels for Preventing Implant-Relat

15 בדצמבר 2024, 14:00

חדר סמינרים -חדר 315

ד"ר גלי פיכמן-Bio inspired Design of Peptide-Baseded Infections Functional Gels for Preventing Implant-Relat

Abstract: Self-assembling peptides offer great potential as versatile building blocks for developing functional hydrogels suitable for various applications.

This presentation will focus on the mussel inspired design of peptide-based antibacterial hydrogels.

These bioinspired materials combat bacteria through a dual mechanism involving surface-contact membrane disruption and oxidative killing facilitated by hydrogen peroxide (H2O2) production. The leading formulation, MIKA2, has shown its potential as an injectable hydrogel capable of preventing implant-related bacterial infections, effectively reducing bacterial colonization on titanium implants in mice.

LMI Seminar: Time-resolved attosecond interferometry

04 בדצמבר 2024, 13:00

הפקולטה להנדסה אוניברסיטת תל אביב, בנין כיתות ,אולם 011

LMI Seminar: Time-resolved attosecond interferometry

סמינר מחלקתי של ברק הלפרן- מפסולת לערך: טיפול מקדים באוזון כפלטפורמה להשבת משאבים בת קיימא

30 ביוני 2025, 14:00 - 15:00

From Waste to Worth: Ozone Pretreatment as a Platform for Sustainable Resource Recovery

Monday June 30th 2025 at 14:00

Wolfson Building of Mechanical Engineering, Room 206

Abstract:

As global pressures mount around waste accumulation and resource scarcity, advanced oxidation processes- particularly ozone-based treatments- offer promising pathways to transform waste into valuable materials. This seminar presents a series of interdisciplinary investigations into ozone-assisted strategies for upgrading lignocellulosic, textile, and plastic waste streams into functional products.

For lignocellulosic waste, a hydrodynamic venturi reactor was engineered using a “Free model” configuration, allowing unrestricted circulation of particles and achieving nearly 100% ozone utilization. This design improved mass transfer, eliminated the need for off-gas treatment, and significantly enhanced enzymatic hydrolysis yields- up to 65%- while enabling valorization of residual solids into materials suitable for adsorption. This approach supports a scalable, near-zero-waste biorefinery model.

In the treatment of textile waste, ozone was implemented as a post-treatment “polishing” and bleaching step in nanocellulose (NC) production. The process improved NC purity and functional surface properties, while enabling recovery and reuse of sulfuric acid- commonly discarded in conventional protocols- thereby reducing operational costs and environmental hazards.

The third case explores an emerging biotechnological frontier: Plastic waste valorization explored the integration of ozone with fungal biodegradation. Ozone pretreatment increased the hydrophilicity and surface reactivity of nylon, promoting colonization by Trametes versicolor. The resulting fungal biomass shows promise as a protein-rich additive for food or feed, presenting an innovative route for transforming recalcitrant polymers into nutritive resources.

Additional innovations include reactor design optimization for energy efficiency, streaming potential surface characterization, and ozone application for pollutant degradation. Together, these findings demonstrate that ozone pretreatment, when coupled with tailored bioprocessing, provides a flexible and scalable platform to advance circular economy goals, integrating waste management, biotechnology, and sustainable engineering.

Bio:

Barak Halpern is a multidisciplinary environmental engineer and biotechnologist specializing in advanced oxidation processes for waste valorization. His doctoral research at Tel Aviv University, conducted in collaboration with the University of Toronto, Oranim College, and Kinneret College, focuses on ozone-based pretreatment technologies to enhance the biodegradation and upcycling of organic and plastic waste. His work integrates reactor engineering, biotechnology, and circular economy principles to convert waste streams.

A hallmark of his Ph.D. research is the design of an innovative hydrodynamic venturi reactor operating in a "Free model" configuration, which achieves nearly 100% ozone utilization and significantly improves enzymatic hydrolysis efficiency. He has demonstrated the reactor’s scalability in a semi-pilot biorefinery producing disinfectant-grade ethanol during the COVID-19 pandemic. In related projects, Barak has developed processes for recovering and reusing sulfuric acid in nanocellulose production and for accelerating fungal degradation of ozone-treated plastic waste.

Barak has published in high-impact journals (e.g., Cellulose, International Journal of Molecular Sciences) and presented at leading conferences in sustainability, food tech, and polymer recycling. He has received multiple awards, including the Teaching Commendation for excellence in education (2023/2024), the KKL Climate Scholarship, and 2nd place in the Falling Walls Israel Lab competition for his work on agri-plastic waste biodegradation, and participating on Coller 2025 contest.

Beyond the lab, he is an experienced mentor, educator, and social entrepreneur, having led sustainability-focused engineering projects through Engineers Without Borders and student-led initiatives addressing pollution, recycling, and environmental resilience.