Connect with us

Deep Learning

Can A Deepfake Be The Achilles Heel Of IPhone Security?

Published

on

Can A Deepfake Be The Achilles Heel Of IPhone Security?

Having a face ID locking system for your phone is the safest way of keeping it secure. Or you think so. Security experts attending a black hat convention in Las Vegas in 2019 bypassed iPhone ID in just 120 seconds. All that researchers needed was a pair of spectacles, a tape, and, of course, a sleeping or unconscious iPhone user. They could achieve this because of the flaw they found in the liveness detection function of the biometric authentication system that is used by Apple for unlocking an iPhone. This was a real WTF moment at the conference. Well, breaking into phone security is not new.

There are many other instances when phone security was hacked using 3D-printed faces. In a series of ‘pretend attacks’ carried out by Sensity AI, a startup company focused on tackling identity fraud, used scanned images of similar images of a person to breach into his phone security. And it worked. Now, with deep fakes turning upside down the belief ‘Seeing is believing’, phones are not safe from unauthentic access.

Hold on. If you are an iPhone user, you can take a sigh of relief. Deepfakes seem to take not so well with the iPhone thanks to the deep sensor technology which the iPhone uses, along with its camera, to look at the face. Here is how it works: the iPhone facial recognition system comes with an infrared dot projector, which when covers the face with invisible points of light, and an infrared camera to form the 3D structure of the face. But remember, it is still vulnerable to 3D masks, so there is no reason to leave your iPhone alone.

Advertisement

The post Can a Deepfake be the Achilles Heel of iPhone Security? appeared first on .

Artificial Intelligence

An Overview Of Deep Reinforcement Learning

Published

on

An Overview Of Deep Reinforcement Learning

Deep reinforcement learning, in which machines may train themselves based on the outcomes of their activities, is one of the most exciting fields of artificial intelligence today. It is one of the most promising fields of artificial intelligence. Read on to learn more about Deep Reinforcement Learning.

What is deep reinforcement learning?

Deep reinforcement learning is an artificial intelligence and machine learning category in which intelligent robots can learn from their behaviors in the same way that humans learn from experience. The fact that an entity is rewarded or penalized based on its activities is inherent in this sort of machine learning. Actions that lead to the desired end are rewarded (reinforced).

A machine learns through trial and error, making this concept perfect for dynamic, ever-changing surroundings. While reinforcement learning has been present for decades, it was only later that it was paired with deep learning, which produced spectacular results.

Applications of Deep Reinforcement Learning

AI toolkits for training

Advertisement

AI toolkits for training

AI toolkits like OpenAI Gym, DeepMind Lab, and Psychlab are offering the training environment required to launch large-scale deep reinforcement learning innovation. These open-source technologies are used to train DRL agents. We will continue to witness tremendous development in practical applications as more organizations employ deep reinforcement learning for their distinct business applications.

Manufacturing

Intelligent robots are increasingly being used in warehouses and fulfillment centers to filter through millions of products and distribute them to the correct recipients. When a robot selects a gadget to place in a container, deep reinforcement learning assists it in learning whether it succeeded or failed. It will make better use of this knowledge in the future.

Automotive

Deep reinforcement learning will be powered by a diversified and huge dataset from the automobile industry. It is already being used for autonomous vehicles and will help alter factories, vehicle maintenance, and total industry automation. The sector is driven by quality, safety, and cost, and DRL will bring new ways to enhance quality, save money, and have a greater safety record by combining data from consumers, dealers, and warranties.

Advertisement

Finance

Its goal is to use artificial intelligence, especially deep reinforcement learning, to be good investment managers than people and to analyze trading methods.

Healthcare

Deep reinforcement learning has enormous potential to transform healthcare, from selecting optimal treatment options and diagnosis to clinical studies, new drug research, and automatic treatment.

Bots

Advertisement

Deep reinforcement learning is used to fuel the conversational UI approach that enables AI bots. Because of deep reinforcement learning, bots are quickly learning the intricacies and semantics of language across many areas for autonomous speech and natural language understanding.

The prospect of deep reinforcement learning has sparked a lot of interest. Since this subset of AI learns by interacting with its surroundings, the possibilities are virtually limitless.

More Trending Stories 
  • What Does it Exactly Take for an AI to Become Perfectly ‘Sentient’?
  • Generative AI to Help Humans Create Hyperreal Population in Metaverse
  • Hertzbleed Hacks Target Computer Chips to Steal Sensitive Data
  • DeepMind’s AlphaFold2 Solves the Long-Stood Protein-Folding Problem
  • Will Sentient AI Gain Equal Rights as Humans in the Future?
  • Crypto Sphere has 20,000 Tokens Now! 10 Best to Buy Today
  • Python is No More Overrated, Thanks to Python 3.11

The post An Overview of Deep Reinforcement Learning appeared first on .

Advertisement
Continue Reading

Deep Learning

Why GFlowNets Are Useful? Talk About How They Are Trained

Published

on

By

Why GFlowNets Are Useful? Talk About How They Are Trained

In this article, we provide a guide to building your first GFlowNets in TensorFlow 2.0 implementation

GFlowNets is an ML technique for generating compositional objects at a frequency proportional to their associated reward. It was introduced at NeurIPS by Emmanuel Bengio. They live somewhere at the intersection of reinforcement learning, deep generative models, and energy-based probabilistic modeling. GFlowNet is very useful in a combinatorial domain, drug molecule synthesis.

Generative Flow Networks are a DL technique for building objects at a frequency proportional to the expected reward of those objects in an environment.  They allow neural nets to model distributions over data structures like graphs, the sample from them as well as estimate all kinds of probabilistic quantities which otherwise look intractable. In this article, we are a guide to building your first Generative Flow Networks in TensorFlow 2.0 implementation.

Generative Flow Networks:

Generative Flow Networks have been introduced as a method to sample a diverse set of candidates in an active learning context, with a training objective that makes them approximately sample in proportion to a given reward function. It aimed at bridging the gap between SOTA AI and human intelligence by introducing system 2 inductive biases in neural nets. They amortize in a single but trained generative pass, the work is typically done by computationally expensive MCMC methods.

Advertisement

GFlowNets was to sample the distribution of trajectories whose probability is proportional to a positive return. It can be used to estimate joint probability distributions and corresponding marginal distributions and are particularly interesting to represent distributions over composite objects like sets and graphs. It is a probabilistic model that builds objects. And it selects one building block at random according to the predicted distribution. GFlowNet can turn a given energy function into samples but it does it in an amortized way, converting the cost of a lot of very expensive MCMC trajectories into the cost training of a generative model.

Flow Networks:

A flow network is a directed graph with sources and sinks and edges carrying some amount of flow between them through intermediate nodes of pipes of water. The motivation of such a flow network is iterative black-box optimization, where the agent has to calculate the reward for a large batch of candidates for each round. Some real-life problems like the flow of liquids through pipes, the current through wires, and the delivery of goods can be modeled using flow networks.

How GFlowNets are trained:

The Flow part of GFlowNets is water flows from the origin, through all the actions, and flows out at terminal states where rewards are known. This goal is to distribute water through the pipes such that it flows out in volumes proportional to the rewards.

Advertisement

Generative Flow Networks will stick with Trajectory Balance. It loss function is one way to achieve that goal. For each trajectory, the forward flow of probabilities. A key point is that the reward and terminal state are fixed values in your training data, but the trajectories are learned.

Build a GFlowNet in TensorFlow 2.0:

TensorFlow 2.0 is a library that provides a comprehensive ecosystem of tools for developers, researchers, and organizations who want to build scalable Machine Learning and Deep Learning applications. While the Bengio lab implemented their GFlowNets using PyTorch, and TensorFlow 2.0. A properly trained GFlowNet generates objects with probability proportional to the reward.

The post Why GFlowNets are Useful? Talk About How They are Trained appeared first on .

Advertisement

Continue Reading

Deep Learning

Deepfake Is The Scariest Thing Happened To Mankind

Published

on

Deepfake Is The Scariest Thing Happened To Mankind

Let’s see how the deepfake scenario is an emerging challenge that has burdened many industries

Digital transformation has opened more avenues for businesses and simultaneously the challenges took shape. The deepfake scenario is an emerging challenge that has burdened many industries in these years.

 

So, let’s understand what are deepfakes?

As the name suggests, these are realistic-looking fake images, videos, and audio that leverage AI and deep learning technology. Deepfakes are created using deep neural networks (DNN) and generative adversarial networks (GAN).

 

Advertisement
A Threat to Media

Fake News: Content is the backbone of media and deepfakes directly affect them. Disinformation and fake news are eating away the credibility and trust in news and media.

Visual Communication: Deepfakes are adversely affecting the authenticity of visual communication by spreading synthetic re-enactment videos.

Social Media: The recent deepfakes of influential politicians like Barrack Obama and Donald Trump created havoc on social media. This reduces the reliability of social media and news platforms.

If wrongly used, deepfakes will reinforce false beliefs and provoke unpleasant actions among the audience, since the media is powerful and influential.

 

Advertisement
A Threat to Politics

The emerging threat of deepfakes could have an unprecedented impact on politics. This AI-powered technology is already starting to threaten democracy, democratic elections, policy-making, and society at large.

Here are major areas where Deepfakes could be a risk:

Election: Deepfake technology could impact election processes by spreading fake news related to government policy, initiatives, etc.

Fictitious Content: Deepfakes can be used to create bogus content on digital platforms, including offensive or controversial statements to incite violence.

Political Advertising: By using deepfakes, bad actors can intentionally produce misinformation. They can deflect political ads made for national benefits.

Advertisement

If not regulated now, deepfakes could bring toxic politics in the future.

The post Deepfake is the Scariest Thing Happened to Mankind appeared first on .

Advertisement
Continue Reading

Trending

Get our daily News updatesSignup to get instant updates straight to your email