Purestform @purestform - Tumblr Blog

Most people make the mistake of thinking design is what it looks like. People think it’s this veneer – that the designers are handed this box and told, ‘Make it look good!’ That’s not what we think design is. It’s not just what it looks like and feels like. Design is how it works.

Steve Jobs

formlab

(via 2010-196786-the-new-amg-5-5-litre-v8-biturbo-engine1.jpg (1920×1080))

look-at-stuff

Airpod Pro I Apple (2019)

Clothing has to make sense with what you are and what you do. I think all the creatives around the world should take a position and reject what doesn’t represent them by working and acting in alternative ways.

Boris Bidjan Saberi

#dark fashion #avantegarde #boris bidjan saberi

searchsystem

Spin / Spin Projects / Copernicus Crater Moon / Typography / 2020

formlab

https://www.press.bmwgroup.com/global/article/detail/T0278839EN/aerodynamic-changes-to-the-bmw-m4-dtm-promise-even-more-excitement-for-the-2018-season

#automotive #bmw

visualtorque

http://sid766.tumblr.com/

The Value Function

Reinforcement learning is about finding the combination of actions that yield the best cumulative reward. To find these actions, it is useful to first identify goals and traps, then assign rewards and penalties to each action with respect to where the goals and traps are at each iteration. The maximum possible reward can be viewed as performing the best action for a given state.

The trained or fitted value function assigns rewards or penalties for each action at every iteration. For example, completing the second iteration is more valuable than the first because the second iteration represents a state which is closer to the goal. The value function allows agents to score actions and select the most rewarding.

In the figure above, the value function is iterated over all states over time. Here each square is a state: S is the start state, G the goal state, T squares are traps, and black squares cannot be entered. In value iteration we initialize the rewards (traps and goal state) and then these reward values spread over time until an equilibrium is reached. Depending on the penalty value on traps and the reward value for the goal different solution patterns might emerge; the last two grids show such solution states.

#artificalintelligence #algorithm