Artificial intelligence (AI) is an area that focuses on enabling machines and software to process information and make decisions autonomously. Machine learning, a component of AI, involves computer systems enhancing their problem-solving and comprehension of complex issues through automated techniques.

The three central machine-learning methodologies that programmers can use are supervised learning, unsupervised learning, and reinforcement learning. For in-depth information on supervised machine learning and reinforcement machine learning, kindly refer to the articles dedicated to them. Here you can read up on the basics of unsupervised machine learning.

Image source: Getty Images.

With unsupervised machine learning, a system is like a curious toddler exploring a world they know nothing about. The system is exploring data without knowing what it's looking for but is excited -- in a digital kind of way -- about any new pattern it stumbles upon.

With this type of machine learning, algorithms sift through heaps of unstructured data without any specific directions or end goals in mind. They are looking for previously unknown patterns, much as you might look for a new stock pick in an overlooked corner of the market. This is rarely the last step since the owner of the raw data typically applies more sophisticated deep learning or supervised machine learning analyses to any potentially interesting patterns.

Why should you care about this artificial intelligence toddler on a quest without a firm goal? Well, unsupervised machine learning is actually on the cutting edge of technology and innovation. Its a key player in everything from autonomous vehicles learning to navigate roads to recommendation algorithms on your favorite streaming platforms. This pattern-finding method is a powerful first step in a deep analysis of any complex topic, from weather forecasting to genetic research.

Two major types of unsupervised learning are clustering and association.

So now you know what unsupervised machine learning is and why it matters. How can you take this newfound knowledge and put it to good use?

First off, you can make informed investment decisions. Companies leveraging unsupervised learning are often poised for growth as this technology continues to evolve. Think about Amazon (AMZN -1.27%) using unsupervised learning for its product recommendations or Netflix (NFLX -2.99%) running unsupervised machine learning routines across years of collected viewership data to generate your streaming home page and make future content production decisions.

These applications aren't just fun toys -- they are business advantages and growth drivers.

Also, AI and machine learning continue to reshape many industries. Whether you're into FAANG stocks or emerging AI startups, knowledge about unsupervised learning can give you an edge in evaluating a company's tech prowess and potential for future success.

We all appreciate a bit of connection, right? Well, thanks to unsupervised machine learning, we're getting better at finding people we might know or like on social media platforms. Facebook is a prime example.

Have you ever wondered how Facebook seems to know who your actual friends from high school are -- the ones you may actually want to keep in touch with? It's not sorcery. It's unsupervised learning in action.

Meta Platforms' (META -0.29%) massive social network continually analyzes a trove of user data, looking for patterns and shared features among users. Common friends are a helpful clue; similar locations and shared interests can point the platform in the right direction, and mutual workplaces can be the clincher. None of these qualities is enough in itself to find that long-lost flame or forgotten friend, but they add up through the power of unsupervised machine-learning algorithms.

So when Facebook suggests "People You May Know," it essentially gives you the output of an unsupervised learning model. The social network isn't just pulling these suggestions out of a digital hat. Each one is the result of a complex analysis of patterns and connections.

View original post here:
What Is Unsupervised Machine Learning? - The Motley Fool

In the following section, we present the learning capabilities of the DRL agent with respect to optimizing an airfoil shape, trained in our custom RL environment. Different objectives for the DRL agent were tested, gathered into three tasks. In Task 1, the environment is initialized with a symmetric NACA0012 airfoil and successive tests were performed in which the agent must (i) maximize the lift-to-drag ratio L/D, (ii) maximize the lift coefficient Cl, (iii) maximize endurance (C^{3/2}_{l}/C_{d}), and (iv) minimize the drag coefficient Cd. In Task 2, the environment is initialized with a high performing airfoil having high lift-to-drag ratio and the agent must maximize this ratio. The goal is to test if the learning process is sensitive to the initial state of the environment and if higher performing airfoils can potentially be produced by the agent. In Task 3, the environment is initialized with this same higher performing airfoil, but has been flipped along the y axis. Under this scenario, we investigate the impact of initializing the environment with a poor performing airfoil on the agent and determine if the agent is able to modify the airfoil shape to recoup a high lift-to-drag ratio. Overall, these tasks demonstrate the learning capabilities of the DRL agent to meet specified aerodynamic objectives.

Since we are interested in evaluating the drag of the agent-produced airfoils, the viscous mode of Xfoil is used. In viscous flow conditions, Xfoil only requires the user to specify a Reynolds number (Re) and an airfoil angle of attack (alpha). In all tasks, the flow conditions specified in Xfoil were kept constant. A zero-degree angle of attack and Reynolds number equal to 106 were selected to define the design point for the flow conditions. The decision to keep the airfoils angle of attack at a fixed position is motivated by the interpretability of the agents policy. A less constrained problem, in which the agent can modify the angle of attack, would significantly increase the design space, leading to less interpretability of the agents actions. Additionally, the angle of attack is chosen to be fixed at zero in order to easily compare the performance of agent-generated shapes with those found in the literature. The Reynolds number was chosen to represent an airfoil shape optimization problem at speeds under the transonic flow regime15. Hence, given the relatively low Re number chosen, the flow is incompressible over the airfoil, although Xfoil does include some compressibility corrections when approaching transonic regimes (Karman-Tsien compressibility correction,43). All airfoils are thus compared at zero angle of attack.

Two parameters relating to the PPO algorithm in Stable Baselines can be set, namely the discount factor (gamma) and the learning rate. The discount factor impacts how important future rewards are to the current state: (gamma = 0) will favor short-term reward whereas (gamma = 1) aims at maximizing the cumulative reward in the long run. The learning rate controls the amount of change brought to the model: it is a hyperparameter tuning the PPO neural network. For the PPO agent, the learning rate must be withing ([5times 10^{-6}, 0.003]). A study of the effects of the discount factor and learning rate on the learning process was conducted. This study shows that optimal results are found when using a discount factor (gamma = 0.99) and learning rate equal to 0.00025.

In building our custom environment, we have set some parameters to limit the generation of unrealistic shapes by the agent. These parameters help take into account structural considerations as well as limit the size of the action space. For instance, we define limits to the thickness of the produced shape. If the generated shape (resulting from the splines represented by the control points) exhibits a thickness over or under a specified limit value, the agent will receive a poor reward. Regarding the action space, we set bounds for the change in thickness and camber. This allows the agent to search in a restricted action space thus eliminating a great number of unconverged shapes resulting from actions bringing changes to the airfoil shape that are too extreme. These parameters are given in Table2. Moreover, the iterations parameter is the number of times Xfoil is allowed to rerun a calculation for a given airfoil in the event the solver does not converge. Having a high iterations number increases the convergence rate of Xfoil but also increases run times.

The environment is initialized with a symmetric airfoil having (L/D = 0), (C_{l} = 0) and (C_{d} = 0.0054) at (alpha = 0) and (Re = 10^{6}). In a first experiment, the agent is tasked with producing the highest lift-to-drag airfoil, starting from the symmetric airfoil. During each experiment, the agent is trained over a total number of iterations (defined as the total timestep parameter), which are broken down into episodes having a given length (defined as the episode length parameter). The DRL agent is updated (i.e., changes are brought to the neural network parameters) every N steps. At the end of an experiment, several results are produced. Figure7a displays the L/D of the airfoil successively modified by the agent at the end of each episode.

Learning curves for max L/D objective starting with a symmetric airfoil.

In Fig.7a, each dot represents the L/D value of the shape at the end of an episode and the blue line represents the L/D running average value over 40 successive episodes. The maximum L/D obtained over all episodes is also displayed. Settings regarding the total number of iterations, episode length and N steps for the experiment are given above the graph. It can be observed from Fig.7a that starting with a low L/D during early episodes, the L/D at the end of an episode increases with the number of episodes. Though significant variance in the L/D at the end of an episode can be seen, with values ranging between (L/D = -30) and (L/D = 138), the average value however increases and stabilizes around (L/D = 100). This increase in L/D suggests that the agent in able to learn the appropriate modifications to bring to the symmetric airfoil resulting in an airfoil having high lift-to-drag ratio. We are also interested in tracking a score over a whole episode. Here, we arbitrarily define this score as the sum of the L/D of each shape produced during an episode. For instance, if an episode is comprised of 20 iterations, the agent will have the opportunity to modify the shape 20 times thus resulting in 20 L/D values. Summing these values corresponds to the score over one episode. If the agent produces a shape that does not converge in the aerodynamic solver, a value of 0 is added to the score, thus penalizing the score over the episode if the agent produces highly unrealistic shapes. The evolution of the score with the number of episodes played is displayed in Fig.7b.

Figure7b shows the significant increase in the average score at end of episode, signaling that the agent is learning the optimal shape modifications. We can then visualize the best produced shape over the training phase in Fig.8.

Agent-produced airfoil shape having highest L/D over training.

In Fig.8, the red dots are the control points accessible to the agent. The blue curve describing the shape is the spline resulting from these control points. It is interesting to observe that the optimal shape produced shares the characteristics of high lift-to-drag ratio airfoils, such as those found on gliders, having high camber and a drooped trailing edge. Finally, we run the trained agent on the environment over one episode and observe the generated shapes in Fig.9. Starting from the symmetric airfoil, we can notice the clear set of actions taken by the agent to modify the shape to increase L/D. The experiment detailed above was repeated by varying total timesteps, episode lengths and N steps.

Trained agent modifies shape to produce high L/D.

We then proceed to train the agent under different objectives: maximize Cl, maximize endurance and minimize Cd. Associated learning curves and modified shapes can be found in Figures10, 11, 12, 13.

Learning curves for max Cl objective starting with a symmetric airfoil.

Learning curves for max (C^{3/2}_{l}/C_{d}) objective starting with a symmetric airfoil.

For the minimization of Cd objective, the environment is initialized with a symmetric airfoil having Cd = 0.0341. This change in initial airfoil, compared to the previously used NACA0012 is justified by enhanced learning visualizations.

Learning curves for min Cd objective starting with a symmetric airfoil.

Trained agent modifies shape to produce low Cd starting with a low-performance airfoil.

Similarly, the results show a clear learning curve during which both the metric of interest and the score at end of episode increase with the number of episodes. The learning process appears to happen within the first 100 episodes as signaled by the rapid increase in the score and then plateaus, oscillating around an average score value.

A second set of experiments was performed to assess the impact of the initial shape. The environment is initialized with a high performing airfoil (i.e., having a relatively high lift-to-drag ratio) and the agent is tasked with bringing further improvement to this airfoil. We chose this airfoil by investigating the UIUC database41 and selected the airfoil having the highest L/D. This corresponds to the Eppler 58 airfoil (e58-il) having (L/D = 160) at (alpha = 0) and (Re = 10^{6}), displayed in Fig.14. Results for this experiments are displayed in Fig.15.

Eppler 58 high lift-to-drag ratio airfoil.

Learning curves for max L/D objective starting with a high L/D airfoil.

It is interesting to compare the learning curves and average scores achieved when starting with the symmetric airfoil and the high performance airfoil.

In Fig.16, we can observe that for both initial situations there is an increase in the average score during early episodes followed by stagnation, demonstrating the learning capabilities of the agent. However, the plateaued average score reached is significantly higher when the environment is initialized with the high performance airfoil, given that the environment is initialized in an already high-reward region (through the high-performance airfoil). Additionally, it was observed that a slightly higher maximum L/D value could be achieved when starting with the high lift-to-drag ratio airfoil. Overall, Task 1 and Task 2 emphasize the robustness of the RL agent to successfully converge on high L/D airfoils, regardless of the initial shapes (in both experiments, the agent converges on airfoils having (L/D > 160)). The agent-generated airfoil for Task 2 is represented in Fig.21a.

Initial airfoil impact on the learning curve.

For Task 3, the starting airfoil is a version of the Eppler 58 airfoil that has been flipped around the y axis. As such, the starting airfoil has a lift-to-drag ratio opposite of the Eppler 58 (i.e., (L/D = -160)), thus exhibits low aerodynamic performance. The goal for this task is for the agent to modify the shape into a high performing airfoil, having high L/D.

In Fig.17, we display the learning curves associated to the score and L/D value at the end of each episode when the environment is initialized with the flipped e58 airfoil at the beginning of each episode. A noticeable increase in both the score and L/D values between episode 30 and episode 75 can be observed, followed by a plateau region. This demonstrates that the agent is able to learn the optimal policy to transform the poor performing airfoil into a high performing airfoil by bringing adequate changes to the airfoil shape. The agent then applies this learned optimal policy after episode 100. Moreover, the agent is capable of producing airfoils having lift-to-drag ratios equivalent or higher than the Eppler e58 high-performance airfoil, signaling that the initial airfoil observed by the agent does not impact the optimal policy learned by the agent, but rather only delays its discovery (see Figs.15 and 17).

Score and L/D learning curves when starting with a low performance airfoil.

An example of a high L/D shape produced by the DRL agent when starting with the flipped e58 airfoil is displayed in Fig.18. It is interesting to notice that in this situation, the produced airfoil shares previously observed geometric characteristics, such as high camber and a drooped trailing edge, leading to a high L/D value. The trained agent is then run over one episode length in Fig.19. By successively modifying the airfoil shape, we can observe that the agent is able to recover positive L/D values having started with a low performance airfoil. This demonstrate the correctness of the behavior learned by the agent.

Agent-produced airfoil shape when starting with low performance airfoil.

Trained agent modifies shape to produce high L/D starting with a low-performance airfoil.

Finally, the best produced shapes (i.e., those maximizing the metric of interest) for the different objectives and tasks can now be compared, as illustrated in Figs.20 and21.

Best performing agent-produced shapes under different objectives and a symmetric initial airfoil.

Best performing agent-produced shapes under different objectives and an asymmetric initial airfoil.

The results presented above demonstrate that the number of function evaluations (i.e., the number of times Xfoil is run and converges on a new shape proposed by the agent) depends on the task at hand. For instance, around 2,000 function evaluations were needed in Task 2, while 4,000 are needed in Task 1 and around 20,000 were required in Task 3. These differences can be explained by the distance that exists between the starting shape and the optimal shape. In other terms, when starting with the low performing airfoil, the agent has to perform a greater number of successful steps to converge on an optimal shape, whereas when starting with an already high-performance airfoil, the agent is close to an optimal shape and requires fewer Xfoil evaluations to converge on an optimal shape. The number of episodes needed to reach an optimal policy, however, appears to be between 100 and 200 episodes across all tasks. Overall, when averaging across all tasks performed in this research, approximately 10,000 function evaluations were needed for the agent to converge on the optimal policy.

Having trained the RL agent on a given aerodynamic task, the designer can then draw physical insight by observing the actions the agent follows to optimize the airfoil shape. From the results presented in this research, it can be observed that high camber around the leading edge and low thickness around the trailing edge are preferred shapes to maximize L/D, given the flow conditions used here. Observing the various policies corresponding to different aerodynamic tasks, the designer can then make tradeoffs between the different aerodynamic metrics to optimize. Multi-point optimization can be achieved by including in the reward multiple aerodynamic objectives. For example, if the designer seeks to optimize both L/D and Cl, a new definition of the reward could be: (r = (L/D_{current} + Cl_{current})-(L/D_{previous} + Cl_{previous})) (after having normalized L/D and Cl). However, multi-point optimization will decrease interpretability of the agents actions. By introducing multiple objectives in the agents reward, it will become more difficult for the designer to draw insight from shape changes and link those changes to maximizing a specific aerodynamic objective.

The proposed methodology enables to reduce computational costs by leveraging a data-driven approach. Having learned an optimal policy for a given aerodynamic objective, the agent can be used to optimize new shapes, without having to restart the whole optimization process. More specifically, this approach can be used to alleviate the computational burden of problems requiring high-fidelity solvers (when RANS or compressibility are required). For these problems, the DRL agent can quickly find a first optimal solution, using a low-fidelity solver. The solution can then be refined using a higher-fidelity solver and a traditional optimizer. In other words, DRL is used in this context to extract prior experience to speed up the high-fidelity optimization. As such, our approach can speed up the airfoil optimization process by very rapidly offering an initial optimal solution. Similarly to8, our approach can also be used directly for high-fidelity models. To accelerate convergence speeds, the DRL agent is first trained using a low-fidelity solver in order to rapidly learn an optimal policy. The agent is then deployed using a high-fidelity solver. In doing so this approach (i) reduces computational cost by shifting from a low to a high-fidelity solver to speed up the learning process, (ii) is data-efficient as the policy learned by the agent can then be followed for any comparable problem and, (iii) bears some generative capabilities as it does not require any user-provided data.

As reinforcement learning does not rely on any provided database, no preconception of what a good airfoil shape should look like is available to the agent. This results in added design freedom leading the agent to occasionally generate airfoil shapes that can be viewed as unusual to the aerodynamicists eye. In Fig.22, we compare agent-produced shapes to existing airfoils in literature. The focus is not on the agents ability to produce a specific shape for given flow conditions and aerodynamic targets, but rather to illustrate the geometric similarities found on both existing airfoils and artificially-generated shapes. A strong resemblance between the agent-generated and existing airfoils can be observed. This highlights the rationality of the policy learned by the agent: having no preexisting knowledge on fluid mechanics or airfoils, an intelligent agent trained in the presented custom RL environment can generate realistic airfoil shapes.

We compare five existing airfoils to our agent-produced shapes in Fig.22. In Fig.22a and b, we compare the agent-produced shape to Whitcombs supercritical airfoil. The shared flat upper surface, cambered rear and blunt trailing edge can be noticed51. We then compare agent-generated shapes to existing high-lift airfoils. Here also, the geometric resemblance is noticeable, notably the shared high camber.

Airfoil shape comparison between agent-produced shapes and existing airfoils.

Detrimental effects of large episode lengths.

One observation was made when noticing drastic decreases in the average score at the end of episode after a first period of increase. We believe this can be explained by the fact that when the episode length is large, once the agent has learned a policy allowing to quickly (under relatively few iterations) attain high L/D values, the average score will then decrease because the agent reaches the optimal shape before the end of the episode. Within the remaining iterations before the episode ends, the agent continues to modify the shape hoping for higher performance, but reaches a limit where the shape is too extreme for the aerodynamic solver to converge, resulting in a poor reward. This would explain why we can observe on Fig.23 a rapid increase in the score between 0 and 25 episodes, during which the agent explores various shapes and estimates an optimal policy, and a strong decrease in the score following this peak during which the agent follows the determined optimal policy and reaches optimal shapes before the episode ends.

The results presented above demonstrate the ability of a DRL agent to learn how to optimize airfoil shapes, provided a custom RL environment to interact with. We now compare this approach to a classical simplex method, under the same possible action conditions: starting from a symmetric airfoil, the optimizer must successively modify the shape by changing thickness and camber at selected x positions to achieve the highest performing airfoil in terms of L/D.

Here, the optimizer is based on the Nelder-Mead simplex algorithm, capable of finding the minimum of a multivariate function without having to calculate the first or second derivatives52. In this case, the function maps a 3-set of actions, being [select x position, change thickness, change camber] to a -L/D value. More specifically, taking the 3-set of actions as inputs, the function modifies the airfoil accordingly, evaluates the modified airfoil in Xfoil and outputs the associated -L/D. As the optimizer tries to minimize the- -L/D value, it searches for the 3-set that will maximize L/D. Once the optimizer finds the optimal 3-set of actions, the airfoil shape is modified accordingly and the optimizer is rerun on this new modified shape. This defines what we call one optimization cycle. Hence, the optimizer is tasked with the exact same optimization problem as the DRL agent: optimizing the airfoil shape to reach the highest L/D value possible by successively modifying the shape. During each optimization cycle, the optimizer evaluates the function a certain number of times. In Fig.24, we monitor the increase in L/D with the number of function evaluations.

Simplex method approachL/D increase with function evaluations for different starting points.

In the three situations displayed, it can be observed that the value of L/D increases with the number of function evaluations. However, the converged L/D value is significantly lower than values obtained through the DRL approach. For instance, even after 500 optimization cycles (i.e., 500 shape modifications and over 30,000 function evaluations), the optimizer is unable to generate an airfoil having L/D over 70. We know that this value of L/D is not a global optimum, as an L/D of at least 160 can be reached with the Eppler 58 airfoil from the UIUC database41. Thus, it seems that the simplex algorithm has converged on a local minimum. Furthermore, as demonstrated in Fig.24a and c, the converged L/D value found by the optimizer is highly dependent on the initial point. The airfoil shapes generated using the simplex method can be found in Fig.25.

Gradient-free approach generated airfoil shapes.

In Table3, we compare the converged L/D values, number of iterations and run times of the simplex method and DRL approach. In both approaches, the agent or optimizer can modify the airfoil 60 times. Although the number of iterations and run time are lower for the simplex method, the converged L/D value is far lower compared to the DRL approach.

This rapid simplex approach to the airfoil shape optimization problem highlights the benefits and capabilities of the presented DRL approach. First, the DRL approach seems less prone to convergence on local minima, as very high values of L/D can be achieved. Second, once the DRL agent has learned the optimal policy during a training period, it can be applied directly to any new situation whereas the simplex approach will require a whole optimization process for each new scenario encountered.

See the article here:
A reinforcement learning approach to airfoil shape optimization ... - Nature.com

Machine learning has made significant strides in recent years, demonstrating remarkable capabilities in various domains such as image recognition, natural language processing, and recommendation systems. However, a fundamental limitation of traditional machine learning approaches is their reliance on labeled training data. This requirement poses a challenge when confronted with new, unseen classes or categories. Zero-Shot Learning (ZSL) emerges as a powerful technique that addresses this limitation, enabling machines to learn and generalise from previously unseen data with astonishing accuracy.

Zero-Shot Learning is an approach within machine learning that enables models to recognise and classify new instances without explicit training on those specific instances. In other words, it empowers machines to understand and identify objects or concepts they have never encountered before. Traditional machine learning models heavily rely on labeled training data, where each class or category is explicitly defined and represented. However, in real-world scenarios, it is impractical and time-consuming to label every possible class.

ZSL leverages the power of semantic relationships and attribute-based representations to bridge the gap between seen and unseen classes. Instead of relying solely on labeled training examples, ZSL incorporates additional information such as textual descriptions, attributes, or class hierarchies to learn a more generalised representation of the data. This allows the model to make accurate predictions even for novel or previously unseen classes.

Zero-Shot Learning operates on the premise of transferring knowledge learned from seen classes to unseen ones. The process typically involves the following steps:

Dataset Preparation: A dataset is created, containing labeled examples of seen classes and auxiliary information describing the unseen classes. This auxiliary information could be textual descriptions, attribute vectors, or semantic embeddings.

Feature Extraction: The model extracts meaningful features from the labeled data, learning to associate visual or textual representations with class labels. This step is crucial in building a robust and discriminative representation of the data.

Semantic Embedding: The auxiliary information for unseen classes is mapped into a common semantic space. This step enables the model to compare and relate the features of seen and unseen classes, even without explicit training examples.

Knowledge Transfer: The model leverages the learned features and semantic relationships to make predictions on unseen classes. By understanding the shared attributes or semantic characteristics, the model can generalise its knowledge to recognise and classify previously unseen instances accurately.

Zero-Shot Learning offers several advantages and opens up new possibilities in the field of machine learning:

Scalability: ZSL eliminates the need for retraining models every time a new class is introduced. This makes the learning process more efficient and scalable, as the model can seamlessly adapt to novel categories without requiring additional labeled examples.

Flexibility: ZSL allows for the incorporation of diverse sources of information, such as textual descriptions or attribute vectors, enabling models to generalise across different modalities. This flexibility broadens the applicability of machine learning in domains where explicit training data may be scarce or costly to obtain.

Real-World Relevance: In many real-world scenarios, new classes continuously emerge or evolve. Zero-Shot Learning equips models with the ability to adapt and recognise novel instances, making them more applicable in dynamic environments where traditional models would struggle.

Transfer Learning: ZSL leverages the knowledge gained from seen classes to make predictions on unseen classes. This ability to transfer knowledge opens up possibilities for transferring models trained on one domain to another related domain, even if the new domain lacks labeled examples.

The applications of Zero-Shot Learning are far-reaching and have the potential to transform various industries. Some notable applications include:

Object recognition and image classification in domains where new classes emerge frequently, such as wildlife conservation or fashion industry.

Natural language processing tasks like text categorisation or sentiment analysis, where new topics or categories continuously emerge.

Recommendation systems, where ZSL can enable personalised recommendations for previously unseen items or niche categories.

While Zero-Shot Learning has shown remarkable promise, there are still challenges that researchers and practitioners aim to address. Some of the key areas of focus include:

Semantic Gap: Bridging the semantic gap between seen and unseen classes remains a challenge. Developing more accurate and robust methods for mapping semantic information to feature representations is essential for improving ZSL performance.

Fine-Grained Learning: Zero-Shot Learning is particularly challenging in fine-grained domains where subtle differences exist between similar classes. Developing techniques that can capture and discriminate these fine-grained details is an ongoing research area.

Data Bias: Ensuring the fairness and generalisation of Zero-Shot Learning models is crucial. Models must be designed to handle data biases and prevent biased predictions when dealing with unseen classes.

As research continues in these areas, Zero-Shot Learning will likely continue to evolve, pushing the boundaries of machine learning and enabling machines to learn and generalise from previously unseen data in even more sophisticated ways.

Zero-Shot Learning represents a significant advancement in the field of machine learning by overcoming the limitations of traditional approaches. By leveraging auxiliary information and semantic relationships, ZSL enables machines to recognize and classify novel classes accurately, without the need for explicit training examples. With its scalability, flexibility, and real-world relevance, Zero-Shot Learning opens up new opportunities for applications in various domains. As research progresses and the challenges are addressed, ZSL is set to revolutionise the way machines learn and adapt, paving the way for more intelligent and capable systems

See more here:
Zero-Shot Learning Demystified: Unveiling the Future of AI in ... - YourStory

Artificial intelligence (AI) and machine learning (ML) have been hailed as revolutionary technologies that will reshape industries, improve productivity, and enhance our daily lives. However, behind the AI veil lies a hidden cost that is often overlooked: the energy intensity of machine learning. As AI and ML applications continue to grow, so does the demand for computational power, leading to an increase in energy consumption and environmental impact. This article will delve into the energy intensity of machine learning and explore the implications of this growing concern.

Machine learning, a subset of AI, involves training algorithms to learn from data and make predictions or decisions. This process requires vast amounts of computational power, particularly for deep learning models, which use artificial neural networks to mimic the human brains decision-making process. Training these models can take days, weeks, or even months, depending on the complexity of the task and the size of the dataset. During this time, the energy consumption of the computers running these algorithms can be immense.

One of the most striking examples of the energy intensity of machine learning is the training of large-scale language models like OpenAIs GPT-3. GPT-3, which has been described as one of the most powerful language models ever created, consists of 175 billion parameters and required hundreds of powerful GPUs to train. According to a study by researchers at the University of Massachusetts Amherst, training a single large-scale AI model like GPT-3 can generate as much carbon emissions as five cars over their entire lifetimes, including manufacturing and fuel consumption.

The energy intensity of machine learning is not only an environmental concern but also a barrier to entry for smaller organizations and researchers. The cost of training large-scale models can be prohibitive, with some estimates suggesting that training GPT-3 could cost around $4.6 million in electricity alone. This creates a competitive advantage for large tech companies with deep pockets, potentially stifling innovation and exacerbating existing inequalities in the AI research community.

To address the energy intensity of machine learning, researchers and industry leaders are exploring various strategies. One approach is to develop more energy-efficient hardware, such as specialized AI chips that can perform complex calculations with less power. Companies like Google, NVIDIA, and Graphcore are at the forefront of this effort, developing custom chips designed specifically for AI and ML workloads.

Another strategy is to improve the efficiency of machine learning algorithms themselves. Researchers are exploring techniques such as pruning, quantization, and knowledge distillation, which can reduce the computational complexity of models without sacrificing performance. These techniques can help make AI models more accessible to a wider range of users and reduce the overall energy consumption of the machine learning ecosystem.

In addition to these technical solutions, there is a growing awareness of the need for more sustainable AI practices. This includes considering the environmental impact of AI research and development, as well as incorporating sustainability metrics into the evaluation of AI systems. Organizations like the Partnership on AI and the AI for Good Foundation are working to promote responsible AI development and ensure that the benefits of AI are shared broadly across society.

In conclusion, the energy intensity of machine learning is a critical issue that must be addressed as AI and ML technologies continue to advance. By developing more energy-efficient hardware, improving the efficiency of algorithms, and promoting sustainable AI practices, the AI research community can help mitigate the environmental impact of machine learning and ensure that these transformative technologies are accessible to all. As we continue to push the boundaries of AI and ML, it is essential that we also consider the hidden costs behind the AI veil and work towards a more sustainable future for our planet.

View original post here:
Behind the AI Veil: The Energy Intensity of Machine Learning - EnergyPortal.eu

Applied UV, Inc.

Expected by Q4 2023, PURONet and Airocide to Help Address Estimated $400 Billion a Year Post Harvest Crop Lost Due to Spoilage Along Supply Chain

MOUNT VERNON, NY , June 15, 2023 (GLOBE NEWSWIRE) -- via NewMediaWire Applied UV, Inc. (Nasdaq: AUVI) (Applied UV or the Company), a leading provider of advanced food security and air and surface disinfection technology, has engaged Quantiva, a provider of AI-driven consulting and software engineering services and solutions, to support the further development and expansion of the Companys proprietary indoor air quality monitoring software, PURONet.

The initial objective of the collaboration is to leverage Machine Learning and Artificial Intelligence to expand the capabilities of Applied UVs flagship brand, Airocide, a patented air purification technology originally co-developed by the University of Wisconsin for use by NASA to grow crops in space. The Airocide proprietary technology eliminates airborne pathogens including bacteria, viruses, pollen, mold, yeast, allergens, VOCs, and odors and is known globally for its ability to remove ethylene by utilizing Photo-catalytic Oxidation. It is currently used extensively in the post-harvest food preservation supply-chain with global end users such as Delmonte, Whole Foods and Esmerelda Farms to name a few. According to the Food and Agriculture Organization of the United Nations, there is a world-wide estimated $400 billion per year a post-harvest crop loss due to Spoilage along the supply chain.

Quantiva is working with Applied UV to harness the power of diverse data inputs and Artificial Intelligence for integration into both the Companys existing PURONet indoor air monitoring and control software, and the Companys Airocide product line. This more advanced product offering is expected to improve outcomes in the logistical supply chains of growers to distributors, transportation companies, and grocers; significantly reducing the magnitude of food spoilage and loss which adversely impacts countries, companies, and consumers around the world.

Story continues

Applied UV Founder, CEO and Director Max Munn, stated, We are excited about the opportunity to collaborate with Quantiva to advance the development of our proprietary PURONet indoor air monitoring software, particularly in the rapidly growing field of food preservation and food security. The impact of crop loss is felt all around the world, and we believe our proven Airocide brand can help mitigate this loss, improving societal outcomes globally. We look forward to our next-generation product launch which we anticipate being completed in late 2023.

According to Quativa Co-Founder and Managing Director Tammo Mueller, Our quantitative approach to decision-making involves defining and analyzing opportunities and challenges and delivering effective solutions through a rational, systematic, and scientific process, based on data, facts, and logic. The opportunity to address this global post-harvest problem is at the very core of our values, improving societal outcomes. We look forward to working closely with the team at Applied UV.

In a March 2023 presentation, Maximo Torero Cullen, Chief Economist of the United Nations Food and Agriculture Organization (FAO), detailed that approximately $400 billion of food is lost each year between harvest and retail. Moreover, a staggering 31.15% of all fruits and vegetables are lost globally due to high perishability and a lack of adequate intervention strategies. The majority of these perennial crop losses are due to spoilage and the effects ethylene has in the degradation of high value crops all along the supply chain.

Mr. Torero predicted that solutions to the global challenge of food loss and waste will require the deployment of context-appropriate and resource efficient technologies together with innovation down the food chain that leverages the collection and real-time utilization of a broad array of economic and environmental data.

For more information about the presentation, click on the following link: https://issuu.com/horticulturaposcosecha/docs/maximo_torero_cullen_current_status_of_food_loss_a

About Quantiva

Quantiva empowers businesses with groundbreaking AI-driven solutions and strategic product development, facilitating innovative growth and agility in regulated environments. Quantivas team of business-focused technologists have provided consulting and software engineering services for a long list of companies including such names as Institutional Shareholder Services (ISS), Siemens, UP Medical, Quantitative Radiology Solutions, and The United Nation. Quantiva is a hands-on, full product lifecycle technology consultancy & digital transformation solution provider with expertise in delivering innovation strategies tailored for multiple industries. Quantivas business category focus includes Banking/Finance, Commerce, Healthcare, Media Ownership & Royalty solutions, Supply Chain Management, and Critical Infrastructure solutions. Quantiva partners with clients to formulate business and product strategies all the way to final implementation, to reimagine processes and platforms to maximize efficiency, ensure legal compliance, and deliver maximum ROI. For more information on Quantiva, please visit https://www.quantiva.co/

Follow us on Twitter

About Applied UV

Applied UV, Inc., provides proprietary surface and air pathogen elimination and disinfection technology focused on Improving Indoor Air Quality (IAQ), specialty LED lighting and luxury mirrors and commercial furnishings all of which serves clients globally in both the commercial and retail segments. For information on Applied UV, Inc., and its subsidiaries, please visit https://www.applieduvinc.com.

Forward-Looking Statements

The information contained herein may contain forwardlooking statements. Forwardlooking statements reflect the current view about future events. When used in this press release, the words anticipate, believe, estimate, expect, future, intend, plan, or the negative of these terms and similar expressions, as they relate to us or our management, identify forwardlooking statements. Such statements include, but are not limited to, statements contained in this press release relating to the view of management of Applied UV concerning its business strategy, future operating results and liquidity and capital resources outlook. Forwardlooking statements are based on the Companys current expectations and assumptions regarding its business, the economy and other future conditions. Because forwardlooking statements relate to the future, they are subject to inherent uncertainties, risks and changes in circumstances that are difficult to predict. The Companys actual results may differ materially from those contemplated by the forwardlooking statements. They are neither statements of historical fact nor guarantees of assurance of future performance. We caution you therefore against relying on any of these forwardlooking statements. Factors or events that could cause the Companys actual results to differ may emerge from time to time, and it is not possible for the Company to predict all of them. The Company cannot guarantee future results, levels of activity, performance, or achievements. Except as required by applicable law, including the securities laws of the United States, the Company does not intend to update any of the forwardlooking statements. References and links to websites have been provided as a convenience, and the information contained on such websites is not incorporated by reference into this press release.

For additional Company Information:

Applied UV Inc.Max MunnApplied UV Founder, CEO & Directormax.munn@sterilumen.com

Investor Relations Contact:

TraDigital IR

Kevin McGrath+1-646-418-7002kevin@tradigitalir.com

Read the original here:
Applied UV Retains Quantiva to Integrate AI and Machine Learning Capabilities into Its PURONet and Airocide Systems - Yahoo Finance