ConceptMiners

06/03/2025

KalaKrut Creative is working on developing a comprehensive web platform for artists.

While we are exploring features required for our community services portal, our commitment to protecting our members' data remains a top priority.

To explore data modeling and its potential for our future portal, we have created synthetic data. In previous posts, we highlighted the use of this synthetic data for creating ontologies and knowledge graphs.

Check out our latest Grafana Labs Annotated Treemap dashboard, created with synthetic data, which is grouped by Crypto Currency/Wallet Names, labeled by Genre, and sized by Deals Completed.

Open the link and Click on a rectangle to see the Name, Genre, and Rank of the artist, along with the Number of Deals Completed.

Such meaningful dashboards provide valuable insights to all users and pave the way for better engagements with artists.



🔗https://kalakrutcreative.grafana.net/public-dashboards/75895f554b19418495635e1e3e635556

(Link will expire soon!)

21/02/2025

𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝘁𝗵𝗲 𝗙𝘂𝘁𝘂𝗿𝗲 𝗼𝗳 𝗠𝘂𝘀𝗶𝗰 𝗗𝗮𝘁𝗮 𝘄𝗶𝘁𝗵 𝗞𝗮𝗹𝗮𝗞𝗿𝘂𝘁 𝗖𝗿𝗲𝗮𝘁𝗶𝘃𝗲! 🎶

At KalaKrut Creative, we're passionate about empowering artists and fostering a thriving music community. Leveraging open-source resources and generative AI, we're building a cutting-edge music ontology—a structured way to organize and understand data about artists, genres, collaborations, and more. This is a work in progress, and it will be central to our upcoming Web2/Web3 portal.

𝗪𝗵𝘆 𝗗𝗼𝗲𝘀 𝗜𝘁 𝗠𝗮𝘁𝘁𝗲𝗿?

Currently, industry data is concentrated in the hands of a few major players. A domain-specific ontology for a community-based music initiative is largely absent from the public domain. Data accessibility is a significant challenge. We believe that structured data is key to unlocking the full potential of the music industry. Furthermore, research has shown that AI tool implementations, particularly transformer models, perform more effectively when ontologies are integrated.

𝗢𝘂𝗿 𝗣𝗿𝗼𝗰𝗲𝘀𝘀:

We iteratively created and refined a synthetic dataset using custom Python code, augmented the data with an LSTM model, and developed the ontology using AI completion tools. The resulting ontology was then processed using an RDF grapher tool to generate RDF/XML and Turtle files.

𝗛𝗼𝘄 𝗧𝗵𝗶𝘀 𝗦𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗪𝗶𝗹𝗹 𝗘𝗺𝗽𝗼𝘄𝗲𝗿 𝘁𝗵𝗲 𝗠𝘂𝘀𝗶𝗰 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝘆?

Discover Talent: Quickly identify artists based on genre, location, skills, and even payment preferences.

Facilitate Collaborations: Connect the right people for the right projects by understanding collaboration types and artist availability.

Personalize Opportunities: Tailor opportunities and recommendations to artists based on their unique profiles.

Gain Insights: Analyze trends to help artists make informed career decisions.

Use Cases (In Development):

Enriched Artist Profiles: Showcasing genre, subgenres, location, education, website, social media, discography, NFTs, achievements, collaboration preferences, services offered/sought, deals, posts, engagement metrics, reviews, and rank.

Powerful Search & Discovery: Enabling artists and collaborators to easily connect.

Intelligent Collaboration Matching: Suggesting potential partners based on shared interests and skills.

Personalized Opportunity Recommendations: Delivering relevant opportunities to artists.
Our ontology is in its initial development phase and will be a core component of our future Web2/Web3 portal, evolving as we grow and learn from our community. We're genuinely excited to bring these features to life in future!

05/02/2025

We are pleased to share a bar chart form 's latest and very insightful report.

Avanade Trendlines: Do what matters AI Value Report 2025

The above image reflects varying attitudes towards AI and technology. It could be influenced by many factors, few being eg. how integrated AI is within each country's daily operations as well as media and public discourse especially on AI-related ethics and achievements.

19/06/2024

The image alarms the environmental impact of improperly discarded electronics and the scarcity of elements in our everyday devices. It highlights that a smartphone is made up of over 30 elements, many of which are becoming increasingly rare. The article argues that we should be more mindful of how often we upgrade our phones and how we dispose of them in order to conserve these precious resources.

Legends are self explanatory. Elements used in mobile phones are marked with mobile icon. Serious overusage of elements are in the Red Boxes.

20/11/2023

Visualizations from a report by McKinsey & Company

The state of AI in 2023: Generative AI’s breakout year, August 1, 2023 | Survey

https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023-generative-ais-breakout-year

20/11/2023

Visualizations from a report by McKinsey & Company

The state of AI in 2023: Generative AI’s breakout year, August 1, 2023 | Survey

Link in the first comment.

03/10/2023

Facts are becoming harder to find these days, especially owing to the massive rise in the flow of information on social media, the decline of traditional media, the ever-increasing complexity of issues facing society, and political polarization.

For "fact finding" or "finding the truth", one has to always be skeptical about everything one reads or sees online, as well as perform own fact-checks on the information and source of information while being aware of one's own biases and identifying clickbaits, falsehoods, inaccuracies, omissions of evidence, contexts, references, dates, misinformation and disinformation based on an understanding of the underlying information ecosystem.

Remember, YOU ARE THE SOLUTION

Scholars have developed a classification system for techniques of science denial and the spread of disinformation called the FLICC taxonomy (a chart is included). FLICC stands for fake experts, logical fallacies, impossible expectations, cherry-picking, and conspiracy theories.

24/08/2023

Great Article on Google Research Posted by Wenhao Yu and Fei Xia, Research Scientists, Google, proposes and demonstrates usage of LLMs for Robotic Skills Synthesis through Reward Functions.

Excerpts:

"LLMs struggle to directly output low-level robot commands due to the limited availability of relevant training data. As a result, the expression of these methods are bottlenecked by the breadth of the available primitives, the design of which often requires extensive expert knowledge or massive data collection."

"... reward functions provide an ideal interface for such tasks given their richness in semantics, modularity, and interpretability. They also provide a direct connection to low-level policies through black-box optimization or reinforcement learning (RL)."

"...a language-to-reward system that leverages LLMs to translate natural language user instructions into reward-specifying code and then applies MuJoCo MPC to find optimal low-level robot actions that maximize the generated reward function."

"The language-to-reward system consists of two core components: (1) a Reward Translator, and (2) a Motion Controller. The Reward Translator maps natural language instruction from users to reward functions represented as python code. The Motion Controller optimizes the given reward function using receding horizon optimization to find the optimal low-level robot actions, such as the amount of torque that should be applied to each robot motor."

Examples discuss the use of above system for Robot Dog, Dexterous Manipulator and Validation on Real Robots.

https://ai.googleblog.com/2023/08/language-to-rewards-for-robotic-skill.html

24/08/2023

Great Article on Google Research Posted by Wenhao Yu and Fei Xia, Research Scientists, Google, proposes and demonstrates usage of LLMs for Robotic Skills Synthesis through Reward Functions. Link in the first comment.

Excerpts:

"LLMs struggle to directly output low-level robot commands due to the limited availability of relevant training data. As a result, the expression of these methods are bottlenecked by the breadth of the available primitives, the design of which often requires extensive expert knowledge or massive data collection."

"... reward functions provide an ideal interface for such tasks given their richness in semantics, modularity, and interpretability. They also provide a direct connection to low-level policies through black-box optimization or reinforcement learning (RL)."

"...a language-to-reward system that leverages LLMs to translate natural language user instructions into reward-specifying code and then applies MuJoCo MPC to find optimal low-level robot actions that maximize the generated reward function."

"The language-to-reward system consists of two core components: (1) a Reward Translator, and (2) a Motion Controller. The Reward Translator maps natural language instruction from users to reward functions represented as python code. The Motion Controller optimizes the given reward function using receding horizon optimization to find the optimal low-level robot actions, such as the amount of torque that should be applied to each robot motor."

Examples discuss the use of above system for Robot Dog, Dexterous Manipulator and Validation on Real Robots.

19/08/2023

An image and description is output from a 30 sec music clip. Let us know what do you think about the Llama2 output

https://huggingface.co/spaces/fffiloni/Music-To-Image/discussions/39

A soft, warm light illuminates a cozy living room, with a comfortable armchair invitingly placed in front of a large window. The chair is adorned with a plush cushion in a soft, muted color, and a ...